A commonly held position in social and humanistic psychology is that two mutually distinct response processes are accountable for human behavior. These processes are commonly denoted as extrinsic and intrinsic motivation. However, no attempt has as yet been made to reconcile this position with contemporary single process models of learning that are derived from the empirically grounded disciplines of cognitive science, learning theory, cognitive neuro-physiology, and cognitive linguistics. This paper attempts to reconcile these models through a comparison of their respective parsimony, logical consistency, correlation with observational data or predictive power, and degree to which they may be integrated with other scientific principles and observations. The validity of dual process theories is denied. It is concluded that dual factor theories of human motivation in social psychology do not derive from observational data unique to human behavior, but rather stem from inadequate data languages which can only partially describe the perceptual and neurological facts of experience, and indeed confound, ignore, and needlessly complexify those facts. The economy, comprehensiveness, and applicability of single factor learning theory is demonstrated through an analysis of the ‘flow’ experience, and the philosophical implications of learning theory are noted.
The quintessential question in psychology is how behavior is motivated, reinforced, or otherwise selected. What aspects of our daily experience make a particular behavior more likely than another? In contrast to the highly definable subject matter of the physical sciences, psychology has always been encumbered by a lack of consensus as to the most appropriate way to define the processes that instigate behavior. The sheer difficulty in describing these processes has resulted in differing standards as to what terms may be used to describe how behavior is selected. But different approaches to the description of motivation vary not only in the type of nomenclature or data language used to describe this process, but in how that data language may be applied to a wider range of behavior, and what generalizations or predictions may be derived from that description.
The normative standards that separate good theories from less valuable ones simply reflect the relative capability of a data language to derive the widest scope of events from permutations of the fewest descriptive terms. During this century, and within the discipline of experimental psychology, the essential elements that cause behavior to be selected have been reduced to simple elemental perceptual relationships that may be denoted by equally simple descriptive terms. In contrast, social and humanistic psychology has been characterized by the evolution of ever more complex data languages that have progressively limited generality and predictive power. This eclectic trend in psychology underscores the question of whether human behavior is at all subject to the simplifying trends characteristic of experimental psychology. More pointedly, the question is whether single factor explanations of motivation or reinforcement can be applied without loss to complex human behavior, and if not, whether the origin of human behavior is separated from its animal counterpart due to the evidence of unique mental processes. If complex human behavior can be thoroughly described utilizing the data language provided by contemporary models of reinforcement, then multi process theories of behavior become simply artifacts of a terminology that incompletely or inappropriately describes the subject matters it purports to explain.
A convenient distinction between single and multi process theories of behavior that has great currency in contemporary social psychological thought is the division of the reinforcement process into two separate subject matters of extrinsic and intrinsic motivation. The concept of intrinsic motivation is distinctively illustrated by the concept of the ‘flow response’, a well-demonstrated experiential state of ecstasy, pleasure, or satisfaction that occurs during the performance of tasks that represent the matching of demand and skill. The primary question we will consider is whether the flow response represents a separate definable process that is unique to humans, or if it may be derived from the application of a set of basic principles that are common to both human and animal behavior alike. The answer to this question not only will help define flow, but will also address the viability of theoretical positions which posit distinctive motivational processes, and allow for the equally simple description of other mental processes (e.g. play, hypnosis, stress, etc.) that have been heretofore considered as separate subject matters due to the idiosyncratic data languages which have been developed to describe them.
We will first examine the empirical standards of the physical sciences and note their generality to the very distinctive empirical traditions of modern experimental and social psychology. Secondly, we will compare contemporary motivational models from these disciplines in terms of their logical consistency (syntax), descriptive thoroughness (semantics), applicability to other subject matters (generality), and theoretical economy (parsimony), and will trace the evolution of the concept of reinforcement as it has been conceived in the respective disciplines of experimental and social psychology. Third, we will demonstrate the simplicity and power of contemporary models of reinforcement by deriving complex motivational states such as play, flow, and stress from the permutation of very simple theoretical principles. Fourth, we will examine how knowledge of how behavior is selected informs us as to what behavior we should select. These philosophical implications span such narrow subject matters as the design of games and Internet interfaces to general issues such as ethics, the design of cultures, and the future of the race itself.
Empiricism and Science
Scientific analysis represents the mapping of empirical (sense-data) phenomena to language. Sense data are defined as what we know from our senses and the instruments that enhance them. The ultimate goal of science is a complete description of all of nature, and is essentially a deductive process wherein specific predictions can be reliably made to account for and predict the facts of nature. The object of science is the attempt to move from deductive conclusions that are derived from partial data to inductive conclusions that reflect a complete accounting for all experience. As such, the history of science represents a transition from deductive to inductive knowledge as more and more natural events have surrendered to progressively more comprehensive data languages and procedures. With a more complete accounting for the facts of nature comes a more enhanced ability to predict and control physical phenomena (and vice versa), and to achieve a greater degree of conceptual integration so that diverse events can be seen as aspects of a common pattern (Marr, 1985).
The physical sciences provide perhaps the clearest example of this progression in empirical knowledge. To illustrate, Aristotle deduced the laws of motion from his own observation of physical events, yet his observations were neither systematic nor mapped to a rigorously defined descriptive language (i.e. mathematics). It remained for such scientists such as Kepler, Galileo, and Newton to actually assemble and codify the extensive data that thoroughly described the behavior of matter. The use of these more descriptive data languages provided an increase in their predictive power and generality to many seeming unrelated events. Thus, Newtonian mechanics allowed a scientist to predict accurately not only the behavior of falling balls, but also the flight of feathers and the orbits of moons. Of course, Newton could not account for all the processes which guided the flight of an object in space, and such entities as time, space, and gravity were regarded as unexplained constants that were seemingly barren of any surplus predictive power. It was therefore unnecessary to explain them even though that decision rendered this theory incomplete. Declaring ‘hypotheses non fingo’, or ‘I make no hypotheses’, Newton thought it antithetical to the aims of science to venture an explanation of events that had no correlate in empirical observation. It sufficed merely that the concept of gravity was a concept that was not important to his practical end of predicting the mechanics of motion, and hence merely needed to be noted, not described. Two hundred years later, the concept of gravity did get its due when Einstein demonstrated through his general theory of relativity that gravity was simply the warping of space, and may be described by non-euclidian geometry. Along with a definition of gravity, time and space, Einstein’s theory demonstrated that these entities may be conceptually manipulated to make predictions that would be eventually confirmed by observation and experiment. In such a way, the relativity of time and space were demonstrated, and such bizarre physical entities as time warps and black holes were confidently predicted.
Science represents the continuous attempt to make distinctions between what is essential and what is derived. Historically, the common sense mechanics of Aristotle was subsumed under Newtonian principles, which could be in turn be incorporated under the relativistic principles of Einstein. Each higher level of understanding was represented by a progressively greater degree of generality and predictive power. From a child’s approximation of where a thrown ball will fall to the Newtonian laws of motion to the Einstein view of space time, each successive level of understanding encompassed the lower, and permitted more and more events to be described, new ones to be predicted, and all to be accomplished through the employment of terms of progressively greater economy.
However, an economical description of nature does not necessarily translate into a practical description of nature. That is, describing nature is one thing, but using nature is another, and it may be impractical to use higher order descriptive terms to manage the physical facts of life. Although we can describe the physics of everyday life with Newtonian or even Einsteinian terms, we normally make a trade off from the predictive power and universality of mathematical statements in exchange for simple descriptive terms that have greater utility, but far less predictiveness and generality. Thus, when we throw a ball into the air, we communicate to others not some Newtonian formula, but rather that we threw the ball hard. Of course, the use of a simple description of the act of throwing will not give us the capability to predict exactly where the ball will land, or the ability to use that description to describe how the moon rotates around the earth. But it does give us the ability to generally communicate the motion of throwing a ball, and that is all that is necessary.
We do not of course use higher order mathematics such as the calculus to navigate through life, and our awareness of even the form of Newtonian or Einsteinian interpretations is generally nebulous at best. However, the personal impact of these higher order perspectives is in our awareness of their instrumental value in predicting the motions of a toaster or a star, and how the application of physics through the applied sciences has been instrumental in achieving the material progress of mankind. A less obvious influence of this knowledge is that it constrains us from hypothesizing added terms that do not provide any additional or more precise predictive power. That is, by knowing of Newton or Einstein we can discard Aristotle, and dismiss ideas which would otherwise have the stamp of common sense. Thus the fact that the earth rotates the sun rather than vice versa becomes the new standard of common sense, and the invisible ether is replaced with the bending of space and time.
The progress of the physical sciences represents the evolution of more precise notational schemes that allowed the physical world to be mapped to language, or in other words, to be described. Indeed, the facts of existence are meaningless unless there is a way to codify them, and are merely debatable if they are only partially described. Yet when they are fully described, mathematics coopts argument by proverbially leaving no stone unturned. A fully described event is a fully predictable event, and Newtonian concepts overturned Aristotle because they could accurately describe and thus predict the behavior of matter. Aristotelian notions were squeezed out of the conceptual equation because they could bestow no real or imagined predictive power to events that were already fully predictable. Similar concepts such as phlogiston (a substance that permitted combustion) and ether (a transmission medium for light) were also discarded not because they were unobserved, but because they were ultimately unnecessary.
When information about the world is limited, it invites the postulation of hypothetical processes which fill in the gaps of observation, and which may be used in a post hoc fashion to supplement the limited predictive power of that information. The problem however is that this speculation is immune to logical refutation because there are no logical rules that determine what processes, if any, one is compelled to infer. This results in the creation of competing paradigms that are at least slightly at cross purposes, since neither side will grant the non-empirical assumptions that the other needs in order to make its case, and forces each side to partly talk past each other (Kuhn, 1970). Such speculation is also immune to criticisms of its actual utility because it offers at least the semblance of being marginally useful. Thus, in lieu of any better information, our early ancestors could invent animistic spirits to provide a guiding force behind nature, and could incorporate ghosts and gods in a mechanical chain that at least seems to provide a predictable future. This sort of reasoning explains the persistence of astrology, which fills in the gap of our knowledge of the future by jumping from the correlations of the position of the stars and human behavior to the presumption of causation, and extrapolates these correlations into predictions that have at least the partial semblance of accuracy.
It is the underlying yet central argument of this paper that the great issues of psychology, as in the physical sciences, are ultimately settled not by examining the logic that ties events together, but by improving the language that describes those events. The subject matter of motivation, or of reinforcement, is resolvable not through the constant debate over the relative merits of the logic of derived processes or even of the reliability of the predictions made from those processes. Indeed, this is unprovable or impractical. Rather, the answer can only be found through better data languages that can more comprehensively describe behavior. The question we will address is, are notational schemes available that can more accurately describe behavior, or is behavior inherently indescribable by virtue of its sheer complexity and opaqueness to the descriptive powers of rational introspection and observation?
The Concept of Reinforcement
In keeping with the accepted principles of science, experimental traditions in psychology have always sought the ultimate goal of a complete description of the empirical regularities of behavior that may in turn be enfolded under the most economical set of principles. This compressibility of nature into a spare logical or algorithmic form allows science to derive behavior from a single set of theoretical principles, much like the behavior of the cosmos can be explained by a series of simple mathematical equations. One of the most concise and pivotal explanatory terms in the theoretical lexicon of psychology has been the concept of reinforcement. Reinforcement represents the occasion when behavior is strengthened, directed, and otherwise selected. Reinforcement is by definition a discrete event, yet until recently had not been represented as a directly observable physiological process but was rather inferred from recorded correlations of behavior and other empirical events. As such, reinforcement has been considered a theoretical construct, a mere way of categorizing behavior and its empirical correlates. During this century, the subject matter of experimental psychology has been expanded through the creation of more diverse experimental procedures and more comprehensive data languages to include nearly all of the denotable empirical events that correlate with and consist of behavior. These empirical measures have shifted from a mere recording of overt response measures, such as salivation, bar pressing, etc. to account for all of the empirical relationships that are implicated by behavior. These relationships include not only the sensate properties of physical events or behavior, but the perceptual events which frame them, the neurochemical events that guide the selective perception, and the genetically determined tendencies which cause information to be interpreted or used in certain ways. Thus, perceptual relationships such as time, space, and causality have been included as integral aspects of behavior, and innate or nativistic influences have also been accepted as viable (albeit unobservable) behavioral determinants. That is, behavior can only be understood in terms of the perceptual dimensions it creates, and is defined in terms of future results, present relationships to other events, and historical determinants as revealed through instinct and experience (memory).
In this new interpretation, the content of mental states is understood dynamically by transient rather than static (Port and Van Gelder, 1995) representational properties that may be denoted as brain-environment and body-environment relations. Contrary to traditional symbolic processing approaches to mental states, dynamical mechanisms incorporate temporal events into the representational framework of mental states. The common sense ‘mentalese’ which describes mental states with static and discrete symbolic representations is ultimately inconsistent with the inherently continuous, temporal nature of biological behavior (Treffner, 1997). Thus, the selection of the perceptual content of mental states derives from relational patterns in time (i.e., information) rather than discrete timeless mental attributes. Mental states, and the neuronal assemblies which they represent, may be seen as encoding not only present behavior, but also the temporal contingency of the action required for reaching the goal (Jeannerod, 1994).
This dynamic cognitive model is corroborated by modern learning theory, which along with an expansion of the empirical correlates to behavior has provided an equal expansion of what is meant by behavior, and the derivation of the rules whereby behavior may be selected. According to this new view, behavior is far more than a mere overt response. Rather, behavior represents the ongoing appraisal and selection of information that may or may not be conscious, and may or may not be evidenced in changes in covert or overt somatic responses (muscular flexion, tension, etc.) Reinforcement in turn is the occasion of this selection, and the rules whereby behavior is selected (i.e., the rules of reinforcement) may be applied or mapped to the facts of experience, and thus allow for the more reliable prediction of behavior. The fundamental effect of reinforcement is merely the selection of an environment-behavior relationship rather than the increase of the emission rate of a reinforced response. (Shull, 1995). That is, what is selected is always an environment-behavior relation, never a response alone (Donahoe and Palmer, 1993). This relation can be described as a simple matter of inference (Staddon, 1988), mental rehearsal (Davis and Dougan, 1988), or expectancy (Bolles, 1972). The rules of reinforcement on the other hand simply denote the types of perceptual relationships we will tend to choose and the empirical circumstances that lead us to select them. The types of relationships that we are capable of perceiving and ultimately choose are attributed to evolutionary or nativistic causes. That is, we are genetically predisposed to perceive time, space, and causality (Bolles, 1988), and the manner that we will respond when we perceive certain sets of stimuli (e.g. the smell of a steak, the visual shape of a female, the sound of music, etc.) is attributed to their specific survival value. In addition, we will also choose perceptual relationships that have a general survival value. Thus, perceptual relationships that denote an increase of our power or control over other individuals and things will be chosen over perceptions which do not.
Since reinforcement merely denotes informative relationships, the syntax of reinforcement is therefore simple and logically consistent. Secondly, the more complete mapping of the theoretical principles of reinforcement to empirical observations (semantics) provides greater parsimony due to the possibility to describe many more aspects of behavior with the fewest operating principles. Finally, this new interpretation of reinforcement achieves generality due to its capability to explain other subject matters that had hereto been explained by less economical and descriptive principles.
The generality of a single factor model of reinforcement has been well demonstrated through its applicability to many of the long standing problems of learning such as extinction, stimulus control (habit), response acquisition, etc. (Donahoe and Palmer, 1993). Indeed, the applicability of reinforcement to diverse aspects of behavior implicates the entire subject matter of learning theory, and maintains the strict empirically based standards that define modern behaviorism. However, this hard won generality has not been extended to other subject matters in psychology. In particular, theoretical principles in social and humanistic psychology posit multiple processes to account for human behavior that often have only a partial basis in empirical observation. In general, behaviors that do not have a clear analogue to animal behavior, such as complex emotional behavior ( a sense of guilt, pride, etc.), creative behavior, or complex behavior such as play, leisure activities, etc. have been attributed to multiple processes which are primarily denoted by self reports, and cannot be described through information processing (Goleman, 1996) models. By denying or minimizing the role of information or cognition, self-reports of emotional behavior, play, etc., become mere metaphorical representations of the neurological structures that instigate them and give them form. Thus, many aspects of human behavior are presumed to be representative of unique neurological processes and structures which may be demonstrated or inferred. For example, the expression of emotional behavior can be attributed to ill defined neural processes meditated by brain organelles such as the amygdala or hippocampus, or to inferred processes that have a genetic origin, such as the sociobiologic explanations that attribute altruism, aggression, etc. to instinctive causes. The result of this is a demarcation between explanatory constructs for behavior that trace behavior to discrete, well defined informative events, and those that trace behavior to indiscrete neuro-physiological events that are unmediated by information and are obscurely connected to more diffusely described experiences as relayed by self reports. These issues, and the experimental procedures that are used to explore them have been neatly demarcated into different subject matters that are represented by extrinsic and intrinsic motivation.
Multiple motivational processes refer to conceptually distinct brain states or categories of experience from which behavior ultimately derives. These processes are typically derived from self-reports, which generally reflect inexact metaphorical interpretations of empirical events. However, by referring to them we can account for the facts of experience and also gain some measure of predictability. Thus an individual who has a sense of self esteem, self control, or self-actualization can describe his mental state through those terms, and a knowledge of these internal states can permit him to predict the likelihood of his future behavior. Our intentions, or likelihood to behave, are thus derived from these intrinsic motivational states. The concept of intrinsic intentionality was coined by the philosopher John Searle (1980) as a conceptual counter weight to the depersonalizing reductionist tendencies in psychology. Intrinsic intentionality represents the ‘aboutness’ of our thoughts, beliefs and desires (Dennett, 1987), and our behavior can only be understood and predicted by understanding how it fits under these practically irreducible categories of experience. That is, short of a neurophysiological explanation, there is no higher order of description that can explain intrinsic intentions. The absence of higher order explanations leaves self-reports as the only practicable means test for the exploration of intrinsically motivated states. The identification of intrinsic motivation with subjective experience (Graef, Csikszentmihalyi and Giannino, 1983) rules out objective (neurally grounded) causes for such states, with the result that behavior becomes self sustaining. That is, people engage in an activity because they enjoy that activity, or perceive their action as an end in itself (Kruglanski, 1978). On the other hand, many behaviors can be described quite well through their correlation to discrete events that are easily measured and reproduced. These strictly empirical measures represent extrinsic motivating events, which motivate by representing future goals. The functional properties of extrinsic events are sufficiently described by the verbal representation of those events and not through any expectancies that may be signaled by those events. That is, extrinsic motivating events are sufficiently described by the mere names we give them, and decomposing them to constituent expectancies does not allow one to better predict behavior. Thus a monetary reward does not gain any predictive power if it was decomposed into positive and negative (i.e., dissonant) perceptual events. These combined positive and negative expectancies represent a list of all the things you can buy with money, or the information it mediates as to the inherent desirability of the behavior that preceded it, as when a child infers that reading behavior must be bad if he is paid for doing it.
Intrinsic motivators represent indiscrete processes that are impractical to decompose, and extrinsic motivators represent discrete events that are unnecessary to decompose. In other words, extrinsic and intrinsic motivators are derived from the relative incapability or lack of necessity to provide a more detailed empirical description of the circumstances that elicit behavior. Their use constitutes a shorthand for the true processes that cause behavior to be selected, and not a systematic empirical description of those processes. Intrinsic and extrinsic motivating characteristics or states only suggest the general rather than the exact circumstances that lead to behavior, and leave the true details to the observer to infer from his own limited observation and personal experience. As such they provide only metaphorical as compared to empirical descriptions of behavior. Thus, if we note that a person is behaving creatively and courageously, the individuals we communicate this fact to will define those motivating states in terms of their own observations of the situation that led to that behavior, and in terms of what they consider as creative or courageous from their own analogous experience. Hence communication will always be quick, but at the price of imperfectly representing the truth.
As a matter or practice or philosophy, a descriptive shorthand for behavior may be limited to those events that can be empirically rather than metaphorically described, and this narrow empiricism is represented by behavioristic points of view, which historically ignore events that cannot be demonstrated through experiment and codified in precise terms. If on the other hand this shorthand is applied as well to events that cannot be or have not been fully described, then empirical principles are implicitly rejected, with the result that any terms may be freely inferred, since they need not be completely anchored to observed facts. The postulation of intrinsic motivational states thus allows for the creation of an endless list of new subject matters by virtue of the idiosyncratic data languages that are used to explain them, and not the empirical data which embody them. The arguing point that continues to separate psychologists is whether inquiry should adhere to a narrow empiricism representing a strict inductive approach that effectively limits the types of predictions about behavior to those events that can be described and predicted, or to reject it in favor of a broadly deductive approach which coins new terms with abandon, and yet strays from the predictability of behavior because the terms refer less and less to behavior. In short, The dilemma is whether one should accept incomplete or inaccurate descriptions of behavior.
This debate may only be resolved by reconciling the separate dynamics of inductive and deductive analysis through the derivation of better procedures and data languages that can more completely and accurately investigate and describe behavior. However, a common misconception is that this has not occurred. The popular appraisal of behaviorism is that that it has not grown past the limited procedures and dogma of fifty years ago, and that the ‘dustbowl empiricism’ (Csikszentmihalyi, 1997) of behaviorism has lost its ascendancy in psychology because of its incapability of explaining the hidden world of intrinsically motivated states. Nothing can be further from the truth. As we will see, the modern history of behaviorism is a chronicle of a gradual evolution of an ever-broader array of principles that at present can map the heretofore-indescribable facts of intrinsic motivation with precision and reliability.
The Language of Behavior
The empirical position in psychology, as generally defined, is that our knowledge, personality, behavior and entire psychological makeup is a product of our experience (Bolles, 1988). Nonetheless, as Neal Miller (1959) noted, pure empiricism is a delusion, since "multitudinous events could be observed and an enormous number of relationships could be determined among all of these events, gathering all the facts with no bias from theory is utterly impossible. Scientists are forced to make a drastic selection…. on the basis of explicitly formulated theory." Theories provide the guidelines that determine a scientist’s interest in and interpretation of specific empirical events. Without these preconceived conceptual guidelines, science would be impossible. Necessarily, these guidelines should only be as broad as necessary to encompass the subject matter that is the focus of the theory. Thus Newton’s laws of motion were postulated to explain only the movement of bodies in space, and not the nature of space or time, and no surplus terms were introduced that did not explain, redundantly explained, or expanded beyond that subject matter.
When empirical relationships are discovered which cannot be explained within these guidelines, then a scientist must expand his procedures and data language to account for them, or else define them as representing a different subject matter that is describable by entirely different sets of laws. At the beginning of the century, Ivan Pavlov's research into the physiology of digestion led to observations that could not be accounted by a description of digestion as a mere homeostatic mechanism. Pavlov observed that salivary and other digestive responses were complexly mediated by perceptual or psychic events, and he created an entirely new set of procedures and data language to explore them. Pavlov’s analysis of these ‘psychic secretions’ unwittingly became one of the bases of the field of experimental psychology. In time, the subject matter of psychic secretions provided a set of descriptive terminology which could map not only salivation, but other covert responses that engaged the smooth and striated musculature, and suggested an exact analogue in hypothetical neurological processes which mirrored the Pavlovian language of stimulus and response.
Rendered most simply, Pavlov’s original experiments explored how instinctive or unconditioned responses, such as salivation, could be evoked not only by unconditioned stimuli that the animal is instinctively prepared to respond to, but also to neutral events that occurred in spatial or temporal proximity to those stimuli. By virtue of this pairing, these neutral stimuli could take on the properties of the unconditioned stimulus, and these conditioned stimuli could be utilized to describe events as wide ranging as the evocative power of language and the essential elements of emotion.
The broadening of the applicability or generality of the Pavlovian data language to many more aspects of behavior than its original subject matter of digestive response has occurred with a corresponding broadening in the empirical description of what constitutes a stimulus and a response. Originally, the only important empirical relationship that occurred within Pavlovian or classical conditioning was the creation of a stimulus-response pairing. A stimulus of course creates relationships in time and occupies space, but the dimensionality or perceptual field of a stimulus was discounted because it did not seem to lend any greater predictiveness than the simple description of a stimulus response pair. For example, if the sight of food occurs simultaneously with the sound of a bell, the bell will take on the properties of the unconditioned stimulus or US (sight of food), and therefore will elicit salivation. How well the stimulus actually predicted the occurrence of food did not seem to matter. Yet the notion that the correlations between a stimulus and the likelihood of a response were important was first demonstrated when laboratory animals (Miller, 1962) were observed to respond differentially to how well a conditioned stimulus predicted the onset of food. The fact that predictiveness does matter was decisively confirmed by Rescorla’s (1967) demonstration that the temporal proximity of a conditioned stimulus or CS (e.g. a bell) was not sufficient for classical conditioning, but that the CS also had to predict the occurrence of the US. The conditioned response or CR (e.g. salivation) also embodies a predictive function through its correlation with the imminent onset of the US (e.g. food). However, if the probable onset of food was not merely correlative but contingent upon the occurrence of salivation, then salivation may be considered to change that probability, and thus ‘operate’ on its perceptual environment. The operative as opposed to correlative relationship between behavior and its consequences adds a measure of control by the individual (Staddon, 1988) , and this element of control, whether perceived consciously or unconsciously, represents operant behavior.
The transition from respondent to operant conditioning represents merely an expansion of predictive events from those that an individual passively assimilates to those that he actively controls. Nonetheless, these two types of conditioning have often been perceived as representing certain types of behavior rather than merely certain classes of predictors. Behavior has often been neatly demarcated between involuntary response classes (e.g. emotion, salivation, stress reactions) and voluntary (e.g. walking, talking) that represent different muscle groups (smooth vs. striated musculature) whose actions are mediated by differing neurological structures. However, as Kimble (1970) demonstrated, so called involuntary and voluntary acts are merely artifacts of the type of proprioceptive feedback that is available to control them. For example, if the act of salivation received prosthetic feedback which permitted that behavior to control a far wider variety of events than the onset of food, salivation would assume the properties of a voluntary behavior, and could be easily recreated without the need for the real or imagined evocative stimulus of food. Likewise, when individuals received delayed or intermittent feedback for simple voluntary behaviors like walking, the voluntary control over that behavior is progressively lost.
The inclusion of predictive or cause-effect relationships introduces a cognitive element that supplants the mechanistic reflex-like process that classical conditioning originally suggested, and expands it to include the class of voluntary or operant behaviors. Because a stimulus mediates relationships, and can only be defined through them, these relationships provide a new order of description that has an analogue to informative rather than neurological events. The fact that the behaviors commonly described by classical conditioning were mappable to predictive patterns of information or expectancies rendered them no different than the perceptual relationships between behavior and reward that sustained voluntary or operant behavior. Although no distinct learning processes could be demonstrated to separate involuntary and voluntary response classes (Hearst, 1975), the data languages used to describe them remain quite distinct. They continue to be widely employed because of their applicability to simple problems in learning and their clear correspondence with the simplified explanations derived from common sense. That is, because classical and instrumental conditioning are not processes, but procedures and effects of procedures (Staddon, 1988), they continue to be employed only because of their practical use in experimental inquiry and everyday language. In simple situations mental events seem to be isomorphic to information denoted by experimental contingencies or simple descriptions, and prediction is reliable and precise. But this isomorphism, if true, would thus extend to more complex behavioral situations, which it does not. The fact that the logic of behavior does not correspond to the formal logic of environmental events or contingencies implies that humans are genetically disposed to use and to process information in certain ways, and that the common analogy of the human mind to the relatively straightforward computational capabilities of present day computers is at best imperfect.
Predispositions towards the Use of Information
Our environment is awash in information, and the infinite permutations of the informative events that may be derived from our perceptual world range from the concrete (seeing an apple) to the abstract (calculating the volume of a cube). Ultimately, the reason that we do not spend our days as idiot savants performing rote or otherwise meaningless information processing is that we are genetically constituted to favor events which have some meaning. As products of an evolutionary selection process, this meaning is found in information that has a survival value. The concept of drive has long been used as a way of denoting our sensitivity to information that has such survival value. Put in another way, drives represent a collection of basic preferences in biological regulation, and represent basic value (Damasio, 1994). Drive acts as an intermediary between information and behavior to reflect innate sensitivities to specific stimuli that gain importance or salience as a function of the degree to which we are deprived of them. They are in turn recognized and acted upon due to the physical discomfort of hunger, thirst, anxiety, or boredom. These stimuli may have distinct sensate properties, such as the smell of a steak or the shape of a female, or they may represent abstract perceptual properties such as the relative novelty of a stimulus (exploratory drive) or its causal properties (anticipatory drive). This latter sensitivity to stimulus causality represents sensitivity to stimuli that mark greater degrees of control over our physical or social environment.
Of course, the shape of a reaction to a stimulus must also account for the perceptual field that those stimuli create as well as the memory of the general courses of action that follows past experience. A perceptual field represents the dimensional attributes of a stimulus event in terms of its past history, present dimensional characteristics (i.e. height, width, position relative to other stimuli), and possible future histories. For example, a bird has an instinct or drive to build a nest, and will be sensitive to the stimulus properties that correspond with nesting materials. However, nesting materials are scattered about its environment, and the bird must continually choose different flight patterns to obtain them. The nesting drive determines the goal of behavior, but it cannot determine the course or type of the behavior itself. Behavior is ultimately determined by the perceptual field (where the nesting materials are located, how long they will be there) that permits the animal to choose between alternative course of action. Thus, the bird may walk, fly, or even swim to obtain the materials it needs, and it may travel long, short, or circuitous distances to get to them. The rate and form of this behavior is not determined by instinct, but rather by information about its environment that it processes presently and stores in memory for future utilization. However, this memory store is not only strengthened through direct experience, but is enhanced and strengthened through cognitive rehearsal that occurs apart from experience. In particular, for lower order mammals as well as humans, the modeling or rehearsal of perceptual events not only prepares one to respond immediately, but allows the individual to act without having to sample again the same information in its original environmental setting (Davis and Doogan, 1988). In other words, behavior changes not only as a function of the processing of information occurring in parallel with current changes in environmental events, but is also a function of the preparatory processing of important or salient information that derives from and is committed to memory. Furthermore, this processing does not occur intermittently, but continuously. However, the type and amount of information processed may vary, and may change gradually or in discrete steps. On the behavioral level of description, the selection of a particular environmental behavior relation is defined as reinforcement, which on the neural level causes the neurotransmitter dopamine to be liberated in synaptic clefts between coactive pre- and post synaptic neurons (Donahoe and Palmer, 1993). The functional role of dopamine stabilizes active neural representations in the prefrontal cortex (i.e., attention), and thereby protects goal related delay activity against interfering stimuli, (Durstewitz et al. 1999). Dopamine also labels stimuli with appetitive value, and may provide advance reward information before behavior occurs (Schultz, 1999). Dopamine also mediates the cognitive effects of pleasant feelings that may be denoted by self reports of pleasure, happiness, or satisfaction (Ashby, Isen, & Turken, 1999).
Reinforcement may be defined as a molecular event that represents a change in an environment-behavior relationship, and is isomorphic to corresponding neural events. In addition, reinforcement may also be employed as a molar event, and acts as a taxonomy that classifies information processing choices and levels. Molar reinforcers are signified by perceptual markers that mark the transition between the type, degree, and number of perceptual relationships that are processed in a specific interval of time. These perceptual markers may act as reference points or benchmarks for the sum of continual changes or as demarcation points for discrete changes in environmental-behavior relationships. Thus, these perceptual markers may be defined as discrete events (molecular events), or as indiscrete collections (molar events) of perceptual relationships.
For example, an individual walking to the store for a bouquet of flowers may simultaneously entertain all of the romantic possibilities that may ensue when he presents them to his girlfriend. If however he was uncertain as to whether he could afford the same flowers, he may refrain from his romantic reveries until he finds out whether he can indeed buy the flowers. At that time, he may entertain in rapid succession all of the many romantic implications of his dinner date. The time when the flowers are actually possessed corresponds to the moment when his behavior is ‘reinforced’, yet the figurative implication that his behavior is now somehow ‘glued’ to the reinforcer is a but a semantic convenience. On a molecular level, reinforcement occurs at each momentary instant that the individual shifts his attention to a salient or important perception, whether it is a momentary thought of his girlfriend, or a turn in the road leading to the flower shop. On a molar level, these expectancies are signified by verbal representations that act as taxonomies for clusters of expectancies, or as benchmarks for changes in expectancies. Thus going to the store to buy flowers signifies groups of similar expectancies, and discovering that the flower store is open acts as a benchmark for discrete changes in groups of expectancies. Finally, the level of attention given to any aspect of his behavior will depend upon the number and salience of behavior-event outcomes or expectancies that must be processed in time. For example, walking to the store may involve little processing of new information, but discovering the store open will open new possibilities (i.e. the improved likelihood and implications of a successful date) that will involve the immediate processing of many chains of information. This processing of a high level of salient information may also be distinctly pleasurable, and thus give rise to the impression that taking possession of the flowers per se, and not the expectancies mediated by flowers, imparts some unique pleasurable or reinforcing effect.
The individual’s behavior occurs and is directed by arrays of salient perceptual events that may occur continuously or discretely, and may reflect low to high levels of information processing. Save for an awareness of innate response sensitivities or drives that direct the outcome of behavior, the form of behavior is determined entirely by informative events, and no appeal to any further higher order process is required. However, when the informative properties or semantics of a walk to the flower shop are not distinguished, buying flowers may be instead referred to metaphorical processes that reflect the imposition of other perceptual domains. This reference is necessarily an obscure one, since there are no logical reasons to infer separate processes for such intrinsic and extrinsic motivating states save for our inability to preconceive the informative events that underlie them. The last refuge of this obscurantist position is the fact that extrinsic and intrinsic reinforcing states actually ‘feel’ differently, and that different neurological and somatic processes occur concurrently with these states. These emotional states are normally considered to be separate from the perceptual events that may instigate them, and this dichotomy sustains the notion that the difference between intrinsic and extrinsic reinforcement is more than mere semantics. But is rationality "guided by and swamped by feeling" (Goleman, 1996), or are feelings cognitive, computable, and intelligible (Damasio, 1994), and are therefore integral to rather than separate from the perceptual facts of thought and experience?
Predispositions towards the Source of Information
The integration of the separate response classes of respondent and operant behavior as well as nativistic sensitivities (i.e. drive) under singular cognitive explanatory schemes provides a logical and systematic descriptive language for behavior. However, these models still operate under the common notion that information processing occurs separately from other specialized neurophysiological processes, and that the latter normally interferes with rather than enhances cognition. The notion that specialized neurological and somatic processes are antagonistic or are mere appendages to thought processes contrasts with newer findings which conclude that they are integral to cognition and are in themselves cognitive events, and that decision making would be crippled without them.
The reinforcement of covert response is of course an accepted fact. From the psychosomatic response of an upset stomach that a child may display to avoid school to the daily anxieties that propel and inhibit our behavior, somatic responses daily occur and are reinforced because they indirectly create new relationships that have value. This value however is accentuated not only by the results of behavior, but by the reduction or prolongation of the somatic response itself. For example, a student who leaves an exam room not only delays taking the test, but also escapes the anxiety that surrounded it. Moreover, an individual will often scan the central representation of a repertoire of responses, inhibiting those associated with fear and facilitating those associated with the pleasurable reaction of hope (Mowrer, 1960). Nonetheless, the fact that covert somatic responses mediate expectancies suggests merely that they are incidental and not integral to behavior, and that cognition would occur just as if not more efficiently without them. Given the procedures available to learning theorists, it has not been possible to test this assumption by isolating the human mind from the influence of the somatic responses as particularly evidenced in emotion and generally felt by one’s sense of his body proper. However, this isolation has been recently observed from the entirely different subject matter of neurophysiology.
Behavior seems inextricable from the emotions that may underlie it, but anatomically, the basic or primary emotions arise from brain structures other than those that govern reasoning. If these structures become disassociated due to disease or injury, then cognitive and emotional processes may be analyzed as separate functions. As an information processor, the human brain has been likened to a computer that reaches conclusions by consciously or unconsciously examining and comparing all alternatives. Yet when the neural structures that process such cognitions are isolated from other structures which are responsible for emotion response, decision making accuracy markedly declines (Damasio, 1994). Emotions, it appears, are not extraneous to or disruptive of decision making, but are integral to thought..
This relatively new role for emotion particularly emphasizes the cognitive or perceptual aspect of somatic or bodily states. These somatic changes, and their resulting interpretation as emotion, act to restrict the combinatorial possibilities that reason has to face (De Sousa, 1987) and mark and add value to the myriad choices that we do not time to sequentially examine (Thagard and Barnes, 1996). These appraisal approaches to motivation begin with the assumption that emotions are adaptive, and prepare and motive us to respond adaptively to situations (Kirby and Smith, 1997). This adaptive or functional property of emotions has been extended to included low level or background somatic states that represent the continuous somato-sensory feedback received by the body proper (Damasio, 1994). More specifically, somatic responses that are evidenced in emotions, feelings, or the diverse sensory events which constitute daily experience are processed as information, and may act as markers for perceptual relationships, and also as a booster for continued memory and attention. In other words, a somatic marker is not only an aspect of cognitive function that attenuates conscious cognitive appraisal or deliberation, but also performs a reinforcing function through its incorporation into new expectancies that change behavior.
Somatic markers make cognition more efficient by paring the number of alternatives that will be rationally considered, yet the value of the particular decision is also weighted by the value of the somatic marker itself. That is, although decision making is expedited through a somatic marker, the marker also distorts the value of that decision by virtue of its correlation with that decision. When the perceptual marker is mildly pleasurable or painful, this distortion will be minor, but when the marker is very painful or pleasurable, then the value of behavior can become obviously distorted and detrimental to one’s rational goals.
For example, the defining characteristic of stress is the muscular tension that occurs when we are placed in demanding situations (McGuigan, 1976), and tension therefore originates with the cognitive processing of these perceptual events (McGuigan, 1978). However, in almost all popular interpretations of stress, muscular tension is described as a useless artifact from man’s hunter gatherer past, and although signaled by information, stress is merely a reactive state that provides only a marginal adaptive function. In contrast, a somatic marker interpretation of tension would hold that muscular tension permits one to easily shift from one course of action to another without having to consciously deliberate on all of the available alternatives, and therefore promotes cognitive efficiency and provides functionality by adding value to behavior. Tension, as felt as emotional stress or anxiety, tends to focus attention on threatening stimuli and influences perception to interpret ambiguous stimuli as threatening (Williams et al. 1988, Mineka and Sutton, 1992). It is associated with fast response times, minimal involvement of cognition in decision making, reduced memory consolidation, and promotes quick and crude decision-making. In contrast, muscle relaxation promotes more complex integration of information and the elaboration of existing schemas to accommodate new knowledge (Nugent and Mineka, 1994). In other words, muscular tension or relaxation is originated by as well as mediates information, and controls or determines behavioral outcomes. However, the definition of a somatic marker incorrectly implies that it is static and invariant across time, and does not require a descriptive language that can map changes in the form or topography of a somatic response across time. The provision of such a behavioristic language denotes the informational patterns or contingencies that parallel that response, and its introduction transforms the definition of a somatic maker from a passive undifferentiated response to an operant behavior. However, the functional role of a somatic marker as an operant behavior becomes greatly distorted in situations that create high anxiety, with the result that decision making is significantly impaired. For example, a student may feel test anxiety prior to going to school, and may call in sick or otherwise avoid school because of that anxiety. Because anxiety adds value to the student’s decision to avoid school, it distorts the intellectual or rational order of events that he would normally follow, and is sustained or reinforced through its reduction by the act of avoiding school.
By definition, a somatic marker represents a trade off between cognitive efficiency and cognitive distortion. In general this trade off is favorable to the individual. However, the biasing aspect of a somatic marker is not favorable in extreme situations (e.g. high anxiety), and can be absolutely detrimental to rational functioning. The distorting elements of emotion are ironically the price we pay to think effectively, as the individual instances wherein emotions are harmful are outweighed by the general circumstances in which emotions are absolutely essential to decision making.
Besides somatic changes, the subjective experience of neurological events also influences how we feel and how well we can make decisions. The integration of these events with the somatic changes of the body itself provides a formal definition of emotion. Emotion is the combination of a mental evaluative process, simple or complex, with dispositional responses to that process, mostly toward the body proper, resulting in an emotional state, but also toward the brain itself, resulting in additional brain changes (Damasio, 1994). Thus, although the ‘feel’ of a somatic response as embodied in emotion is purely cognitive and computable, brain changes as represented by neurological activity also have a particular feel, and although they cannot be referred to any specific visceral or neuromuscular state, they are often perceived as pleasurable or painful. Cognitive modes which involve the rapid processing of information are more rewarding than those which entail minimal processing, and we often characterize high levels of information processing as euphoric, hypnotic or trance like states in which awareness of normal bodily states is markedly reduced or lost. The neurological substrate for this subjective interpretation has been attributed to neurochemical activity that seems to mark the importance or salience of environmental stimuli. In particular mesolimbic dopamine (DA) activity has been conceptualized as a reward signal that marks the importance of perceptual events (Horvitz , Stewart, and Jacobs 1997), and promotes the effective processing of afferent signals simultaneously arriving at the midbrain. A cascade of multiple salient perceptual events would presumably accentuate DA activity and facilitate the switching among alternative cognitive perspectives (Ashby, Isen, and Turken, 1999). This neuro-chemical activity would not only facilitate the rapid and efficient focusing of the mind on a wide range of images, but would also be frequently interpreted as highly pleasurable. Preliminary confirmation of this has been provided by neuroimaging studies that demonstrated the increased release of dopamine during activities (a video game) that required sustained shifting of a cognitive set (Koepp, 1998).
In contrast, if events are non salient or aversive, dopamine production will be depressed (Schultz, 1998). The emotional concomitants of these states would perhaps correspond with feelings of elation and depression, and confirm Damasio’s (1994) observation that "The cognitive mode which accompanies a feeling of elation permits the rapid generation of multiple images such that the associative process is richer and associations are made to a larger variety of cues available in the images under scrutiny…. The ensuing wealth promotes ease of inference, which may become over inclusive..." "By contrast the cognitive mode which accompanies sadness is characterized by slowness of image evocation, poorer association in response to fewer cues, narrower and less efficient inferences, over concentration on the same images, usually those which accompanied the negative emotional response. The cognitive state is accompanied by motor inhibition and in general by a reduction in appetite and exploratory behaviors."
Ultimately, reinforcement is dependent not only upon the type and the amount of information processed, but also upon the somatic responses that mediate as well as provide information, and the neurological events which are perceived as various degrees of attention, alertness, and may be felt as pleasurable excitement or euphoria. That is, reinforcement not only represents changes in information, but also the nonconscious processes which mediate that information, and may be felt as pleasurable or painful events. Given these facts a new definition of rewarding or reinforcing events emerges that defines them as perceptual events that often have a felt hedonic component. Stimuli are proposed to function as reinforcers to the extent that, on the behavioral level, they provoke a behavioral discrepancy (i.e. a salient change in information). On the neural level, they cause dopamine to be liberated (Donahoe and Palmer, 1993), and on the neuromuscular level they may be marked by somatic responses (e.g. muscular tension, relaxation) (Damasio, 1994). These neural and somatic markers both mark and add value to behavior, and have a felt hedonic component (i.e. they feel good or bad).
The true nature of these hedonic events cannot be preconceived, since that depends upon language that is grounded on a fine grain description of neural, neuromuscular, and perceptual events. Because of the ambiguous nature of the causes and results of the hedonic element of reinforcement, encoding algorithms may nonconsciously impose on them pre-existing interpretive categories, even if the stimuli "objectively" do not match those categories (Lewicki et al. 1992). The verbal or conscious descriptions of these events may be further distorted through the use of verbal structures that are borrowed from different perceptual experiences or domains. This use of metaphorical description can further distort an objective conceptualization of these events. For example, work anxiety or stress is not due to the job per se, as we could otherwise imagine it, but to the perceptual context in which the work occurs (e.g. impossible deadlines, demanding boss, noisy environment). The origin of stress if further distorted through a metaphorical description that permits us to preconceive stress as being ‘pulled’ by a stressful event, as representing negative energy, or may be hydraulically released by cathartic tantrums or a good back rub.
The goal of isolating the content of mental processes and to simultaneously reduce the philosophical encumbrance of interpretative categories derived from naïve experience and its metaphorical representation (i.e. common sense) implicates the subject matter of cognitive science and its sub-disciplines of learning theory, neuropsychology, and cognitive linguistics. Unfortunately, cognitive science is at odds with sub-disciplines in psychology such as humanistic and social psychology that routinely accept metaphorical mental structures as elemental, with no need to ground them to their neural and neuromuscular correlates. This disembodiment of psychological as well as philosophical concepts permeates the social sciences, and has been comprehensively refuted by modern cognitive linguistics (G. Lakoff and M. Johnson, 1999). In particular, the root motivational mechanisms for human behavior as conceptualized in humanistic and social psychology are accepted because they match our own naïve experience, and the metaphors that we use to explain that experience. These basic mechanisms are metaphorically defined as intrinsic and extrinsic motivation.
Intrinsic Motivation and Cognitive Science
The expansion of the perceptual or informative determinants of behavior to include all somatic and neurological responses, along with the integration of dynamic (i.e. temporal) informative factors with inborn instinctive and biological sensitivities provides a higher order language that can describe all of the facts of behavior. Intrinsic and extrinsic motivating states are ultimately derived from this higher order language, and exist as taxonomies and not separate processes. That is, they are linguistic tools that only generally describe the content of environment-behavior relationships, and are not processes themselves. Ultimately, like the similarly dismissed entities of phlogiston and ether that were once hypothesized as the respective catalytic and transmissive mediums of fire and light, an appeal to motivating states is a mere convenience of language. It describes behavior no better than a child describes the motion of a ball by saying that he threw the ball hard.
Expanding upon this analogy, the relative difficulty in describing dynamic continuous events in terms of discrete motivating states demonstrates how the exclusion of the temporal dimension can result in the multiplication of explanatory terms, and illustrates the difficulties in utilizing a two or three dimensional language to describe a four dimensional space. We can ‘explain’ a ball’s flight in the air by saying that its flight is the result of the force applied to throwing it, and the effects of inertia, wind resistance, and gravity. Although a ball in motion possesses dynamic properties, these explanations are not incorporated into a dynamic language that can describe and predict in time the motion of the ball. That description requires the employment of the Newtonian calculus, and it is the calculus, not a static description of dynamic states, that is used by physicists to explain and predict the motion of balls, moons, and satellites. The dynamic properties of inertia, gravity, and the like may be derived from a higher order descriptive language, just as discrete motivating states may be derived from a higher order descriptive language as afforded by modern cognitive science and learning theory. Human behavior represents non verbal movement or motoric responses (e.g. reflexes) as well as verbally based information (e.g. descriptions of houses, mental states) that historically have been described separately, yet both can be united under a single, uniform, abstract type of representation relying on formal symbol manipulating operations. This discounts any fundamental difference between perceptually based and verbally based information (Jeannerod, M. 1994, Pylyshyn, Z. 1984). Thus, verbal representations of mental images and mental states are derivable from a higher order language that represents the symbolic nature of linguistic relationships (Kosslyn et al. 1979). That is, the mental imagery that is at the core of distinct motivating states is not a genuine phenomenon, but derives from the representational mechanisms that are involved in the processing of information generated by the mind, the body, and the perceived world.
Ultimately, the mentalistic language that posits discontinuous, non interactive mental attributes or states is attractive because behavior can be easily described and categorized, and is used because of its economy or practicality in explaining behavior events, and not because it can accurately map and predict changes in perceptual states. Yet, although we may easily dismiss the notion that common sense physics represents an irreducible description of the mechanics of the world, intrinsic and extrinsic motivation remain qualitatively different (Deci and Ryan, 1985) phenomena. This distinction is assumed not because it refers to different informative or physical events (which it does not), but because it provides a de facto explanation for observations that do not at first seem to be reducible to informative or other causes.
Intrinsic motivation has been most commonly invoked to explain how children respond to the contingent rewards for behavior. It has been repeatedly noted that school children only comply with verbal, monetary, or other rewards as long as those rewards are contingent upon performance, and that the quality, permanence, and subsequent desirability of that performance actually declines because of the contingent application of these extrinsic rewards. The application of an extrinsic reward thus decreases the child’s own appreciation of the intrinsic value of the behavior that is being rewarded, and this appreciation may only be recovered when such rewards are discarded. For example, representative studies have demonstrated that children who are paid to drink an unfamiliar beverage as compared to being merely requested to drink it are more inclined to find the drink less appealing in the future (Birch, Marlin, and Rotter, 1984), and that subjects who are paid to write term papers submit work that is far less creative than if they were motivated by the task alone (Kohn, 1993). In particular, grades have been found to undermine creativity, long term memory, interest in learning, and a preference for challenging tasks (Butler and Nisan, 1986, Grolnick and Ryan, 1987). Without exception, these and other studies that attribute student behavior to intrinsic and extrinsic motivators exclude an analysis of the information that is mediated by the reward, the performance, the environmental setting of the behavior, and how it is interpreted due to the personal history of the individual. The refusal to account and control for the informative elements that are otherwise subsumed under the terms intrinsic and extrinsic reward results in simplistic conclusions that would be absurd if proposed in other than in a journalistic setting. For example, the conclusion that "rewards cause people to lose interest in whatever they were rewarded for doing" (Kohn, 1993) would come as some surprise to art and music historians, who would note that much of artistic creation would not have occurred without the "extrinsic" monetary and social reinforcement of popular and aristocratic patronage.
Indeed, the situations that seemingly demonstrate the counterproductive effects of extrinsic motivators are unremarkable, and may be transposed to our own experience. If we were paid for example to drink a new soft drink, we may think less well of the drink that if we otherwise had freely tried it. Our dislike of the drink is not because of the reward per se, but because of contrary or dissonant information about the drink that is mediated by the reward. Contrariwise, we would rightfully expect a trophy or cash reward to increase rather than decrease the motivation of the recipient, whether that person is a musician, artist, or little leaguer, since such awards generally do not mediate dissonant information. Besides the dissonant information that discourages behavior, information may not be presented that actually encourages specific behavior. Thus, a student who hands in a less than creative term paper is not less creative because a grade decreases his motivation, but rather because a grade is not specific to creative behavior. Similarly, a stockbroker who reaches quota by misrepresenting his company is not a poor performer because selling is rewarded, but because those rewards are not specific to selling ethically. The imposition of an award, whether it be a grade for a paper or meeting quota for a stockbroker, will not change behavior unless the teacher or manager makes it contingent upon the behavior he implicitly as well explicitly requires. Thus the effective use of extrinsic rewards requires the active appraisal of the information that is being mediated by those rewards, and how the subject may likewise interpret that information.
Finally, the observation that "learning is devalued insofar as it comes to be seen as a prerequisite for receiving the teacher’s approval" (Kohn, 1993) ignores the fact commonly overlooked by educators and psychologists alike that a knowledge of the facts of a subject matter is not coextensive with the information that leads to a liking of that subject matter. That is, learning about a subject, whether it be classical music, cooking skills, or Elizabethan poetry may not entail actually experiencing the informative elements that make such subjects rewarding. For example, if a child’s first exposure to baseball was in a classroom setting that demanded a rote memorization of the rules of baseball, baseball statistics, and the like, it would be no surprise that baseball would subsequently have less interest subsequently to the classroom course. More specifically, the formalized classroom introductions of the humanities dehumanize them by excluding them from the social setting that fostered and embodied them, and thus create a disinterest in the student. The fact that Shakespearean plays were written and debuted for the pleasure of a nutcracking, woman ogling crowd rather than a group of starchy English professors, and that Mozart’s music was a vehicle for royal and popular entertainment, parties, and the Viennese equivalent of the local sock hop underline the fact that the socially detached formality of learning, not the extrinsic rewards that may follow learning, is more likely responsible for the devaluation of an interest in school curricula. Indeed, for a child to actively like to learn, he must first broadly sample the subject matters he will later study in the original social settings that incubated them, as liking will accompany learning only if it first precedes it.
Intrinsic Motivation and Metaphor
Ultimately, the common sense dichotomy between intrinsic and extrinsic motivation can be traced to the simple metaphor, derived from our appraisal of conscious decision making strategies, that motivational processes are discrete events that occur in a serial order. By saying ‘Johnny walks to his room to get a ball’, walking behavior occurs, and is metaphorically ‘glued’ to a reinforcing event (the ball). Similarly, if Johnny learns reading and writing and arithmetic, taught to the tune of a hickory stick, his behavior is likewise motivated by his avoidance of an aversive event (a whipping). This simple paradigm has unfortunately been copied all too well by school systems, with the well documented (Kohn, 1993) result that students lose interest in their curriculum, and are far less creatively involved in the learning process. The metaphor that likens children to passive computers who merely need the right programming to learn engages yet another metaphor, that of the computer. Our metaphorical conception of the mind as computer derives from our conception of computers as devices that process simple chains of information in a serial order. However, our actual minds process information not serially but massively in parallel. The metaphor of mind as computer reflects a preconception of how information it processed, and not merely that it is processed. Yet this metaphor, as well as the common sense metaphor that motivational process are discrete serial events, precludes the postulation of multiple concurrent informative causes for behavior.
Behavior is the result of the concurrent conscious and nonconscious assimilation of information that is received from innumerable somatic, neurological, and sensory sources. Therefore to describe behavior as guided by singular extrinsic motivators makes the a priori assumption that the informative content mediated by these sources is nonexistent, indeterminate, or unimportant. Thus, the qualitative aspects of behavior that cannot be traced to singular extrinsic causes somehow emerge from the behavior itself. Individuals who are intrinsically motivated are rewarded by the behavior itself, since singular extrinsic causes cannot explain them.
Behaviors that particularly involve multiple concurrent informative events such as creativity and play cannot be explained through the serial cause-effect relationships between behaviors and reinforcers, but they can be explained with equal economy if those relationships were simply conceived as plural and concurrent. For example, in the movie ‘Shakespeare in Love’, Will was commissioned to write a play that would pay his bills and keep his patron from being dispatched for his bad debts. ‘Paying the bills’ is of course a perfect ‘extrinsic motivator’ that does not quite explain the poetry and power of the resulting play: Romeo and Juliet. Thus, posterity, and not a few armchair psychologists, would attribute such genius as arising from some inner muse, a motivating spark that comes from the mere act of writing. Of course, the play is not only the thing, but it also must meet the demands of a Queen who likes pratfalls, a girlfriend who swoons with romantic phrases, a crowd expecting its share of sex and violence, actors who demand good parts, getting out a better play than his rival Christopher Marlowe, not to mention the judgement of posterity. Doubtless, Shakespeare had to be aware of the concurrent demands made by dozens of similar ‘extrinsic’ motivators, and ‘Romeo and Juliet’ certainly met all of them. Without the shaping extrinsic motivators provided by dozens of often-contradictory demands, Shakespeare’s unrivalled corpus of work could never have occurred.
Ultimately, genius, and the creativity that guides genius, is a cultural affair. It occurs because crowds and kings and girlfriends and posterity demand it. The requirement to simultaneously meet the constraints of a dozen different demands requires the exercise of a mental calculus that demands an ever shifting attention to a thousand details, and a subjective experience that one would find nothing less than exhilarating. Ironically, to increase intrinsic motivation and the special pleasures that attend to it is to increase the involvement of those cultural forces that embrace complex and demanding goals. In the final analysis, Vienna demands its Beethoven, Florence demands its Leonardo, and London demands its Shakespeare. Genius is inseparable from its times, or in other words, its reinforcers.
Like a pointillist picture emerging from a thousand dots of paint, intrinsic motivation emerges from the effects of a thousand extrinsic demands that are modeled in the mind. By accounting for these concurrent informative causes for behavior, the linear metaphorical conceptions of intrinsic and extrinsic motivation become untenable. Motivation is a function of multiple interlocking patterns of information that are continually in flux. But motivation is more than information, it is activation, and the ability to account for many concurrent patterns of salient information requires activation or alertness to all relevant possibilities. Generally, the somatic and neurological events that comprise this alertness are consciously appraised as incidental to behavior, with little or no influence on behavioral choice. Thus people generally perform actions to achieve ends which are justified by informative outcomes, whether those outcomes reflect the purchase of a car or the earning of a college degree. However, when information is extraordinarily salient, neural and somatic activation can become strongly hedonic, and very obviously acts to determine behavior and one’s subsequent judgement of value. But just as the distinction between intrinsic and extrinsic motivation is a mere artifact of the refusal to account for concurrent informative events, a similar refusal to account for concurrent neurological and somatic events has also created a special category of intrinsic motivator that is just as spurious. This category of intrinsic motivator falls under the rubric of ‘peak experiences’, and has been distinctively labeled as the ‘flow experience’.
Intrinsic Motivation and Flow
This final and perhaps best example of an intrinsic motivating state is described by the unique and highly pleasurable emotional state that occurs concurrently with situations that ironically would seem hardly conducive to anything remotely pleasurable. In the early seventies, the psychologist Mihaly Csikszentmihalyi discovered from the basis of thousands of interviews a common subjective experience of pleasure, interest, and even ecstasy that was derived from activities that perfectly matched one’s skills with the demands for performance. Rock climbers, musicians, chess players, surgeons all reported this "flow" state when their activities possessed this perceptual structure (Csikszentmihalyi, 1977), and subsequently reported a greater interest in these tasks as a result of this experience. Csikszentmihalyi defined the flow response as a "holistic response" or an "optimal state of experience is one in which there is order in consciousness." "This happens when psychic energy, or attention, is invested in realistic goals, and when skills match the opportunity for action." (Csikszentmihalyi, 1990) Flow "provides a sense of discovery, a creative feeling of transporting a person into a new reality. It pushes a person to higher levels of performance, and leads to previously undreamed of states of consciousness. In this growth of self lies the key flow activities." "Potentially negative experiences can be transformed into flow by 1) setting clear goals to strive toward, 2) becoming immersed in the activity chosen, 3) paying attention to what is happening, and 4) learning to enjoy immediate experiences." The flow state has been defined as "an intrinsic reward for participation in an activity (Csikszentmihalyi, 1977), an optimal state that occurs when there is a balance between the perceived challenges of a situation and a person’s skills or capabilities for action (Csikszentmihalyi, 1990), as a centering of attention on a limited stimulus field (Webster, Trevino, and Ryan, 1993), and is characterized by a heightened sense of playfulness (Webster and Martocchio, 1992), self control (Ghani and Deshpande. 1994), increased learning (Canter, Rivers and Storrs, 1985), and increased positive subjective experiences (Csikszentmihalyi, 1977). In keeping with the implied reality of multiple motivational processes, these definitions scarcely refer to information but rather to mind states such as levels of consciousness, attention, or playfulness or states of intentionality such as intrinsic reward, self control, goal setting, etc.
What is immediately distinguishing about almost all commentary on the flow experience is its thoroughgoing metaphorical content. Flow occurs because psychic energy is invested, consciousness is ordered, undreamed states of consciousness are reached, and when we are immersed into activity. Flow doesn’t refer to behavioral, neural, or somatic variables, but to other domains of perceptual experience that reflect disembodied levels of experience. Flow is composed of distinctive rational, perceptual, and emotional domains that follow with each other like the chain of boxes in a flow chart. Thus a demand/skill match is followed by attention that is followed or attended to by ‘senses’ of playfulness, self control, enjoyment, etc. All of these different domains act as different segmented psychological faculties. These domains represent different psychological phenomena, and are seen as separate but interacting agencies. Flow is distinctive among intrinsic motivating processes because it is not only signaled by a perception of a matching of skill and demand, but also incorporates other intrinsically motivating processes such as hypnosis, play, self actualization, and psychic energy (attention). Indeed, the critical element of purely informative or perceptual events such as the matching of skill to demand seem almost incidental when lost in this profuse list of motivational processes which are incorporated into the flow construct.
In contrast to the profusion of interpretive or metaphorical characteristics of flow, the physiological correlates to flow have been scarcely investigated and have been only generally or partially described, or in the case of somatic responses, not described at all. Indeed, the neural correlates to flow have been reduced to metaphorical representations of the mind that engage hydraulic or electromechanical models that render the mind with cartoonish simplicity. Thus, flow is depicted as the result of the vaguely defined containment, channeling and alignment of emotions, feedback loops, attentional energies and forces, mental "cool" states, or as the tuning of the arousal and inhibition of neural circuitry (Goleman, 1996). More remarkably, there has been no experimental investigation into the nature of the somatic states that parallel flow. Specifically, the visceral and musculoskeletal concomitants to flow have never been examined.
However, these musculoskeletal correlates may be inferred from a related experience to flow called meditation. The circumstances which create flow as well as its unique experiential characteristics are generally shared by and may be subsumed under the class of ‘meditative’ experiences that have been studied far more exhaustively. Pleasant emotional experiences that are characteristic of meditation have long been associated with strict attentiveness to specific stimuli or stimulus classes and seen as a byproduct of that attentiveness (Goleman, 1976). Like flow, meditation has been associated with unique physiological and experiential states (Goleman, 1976, Brown, 1977; Deikman, 1963), but unlike flow, the neurophysiological and somatic correlates to meditation have been exhaustively studied Fenwick (1977),Michaels (1976), Wagstaff (1975), Pagano and Warrenburg (1983), Holmes (1984, 1988), and have been found to represent no unique neurophysiological or somatic state, but are merely relaxation. The fact that meditative experiences represent no unique physiological state calls into question the validity of similar claims that flow reflects such states, and refers the topic back to an analysis of the informative events which sustain the state of relaxation which is most likely a major component of the flow response.
The purpose and results of Csikszentmihalyi’s study of flow has been to discover the separate components of flow, and not to derive flow from higher principles or to demonstrate how the terms denoting those components can translate into one another. However, this conceptualization of flow invalidates it as a scientific construct since it fails to meet the major criteria that define acceptable scientific inquiry. The scientific value of the flow concept is marginal at best not because the facts of flow are wrong, but because those facts are not logically interconnected or may be logically derived from higher theoretical principles. That is, "If we accept the crucial assumption that the overall aim of science is to discover the maximum amount of order inherent in any field of inquiry, then paradigms that aim simply at discovering ‘what’s there’ without any commitment to discovering orderly relationships would be then judged unscientific or less scientific than their competitors." (Harris, 1979) The scientific inadequacy of the flow concept derives from its poorly developed logical structure and the resulting difficulties in operationalizing or using flow to predict and explain other facts. Because flow has a poor logical basis, it is difficult to understand through the application of logic, and the difficulty in communicating logically what flow means relegates the concept of flow to an intellectual backwater since it cannot be easily described utilizing related data languages of such subject matters as neurobiology or learning theory.
The decomposition of flow into mere lists of real and hypothetical processes underlines its weaknesses as a scientific construct. Primarily, the theoretical explanation for flow has poor syntactical structure because of the uncertain way its defining terms are related to each other. Flow is generally described as a loosely integrated collection of mental or motivational states or processes that are validated primarily by self reports. Because these states are only described metaphorically through self reports, and can be subsumed under no higher order principles, they cannot logically translate into one another, and therefore cannot be related to one another in any logical way. Secondly, the semantics of flow is weak to non-existent. Outside of the self reports which describe flow experiences, the theoretical terms for flow refer to few behavioral or informative events, with only a fragmentary or metaphorical consideration of its physiological correlates. Thus no aspects of flow can be traced in time, since none are precisely defined. Ultimately, the flow construct is unparsimonious and incomplete since it relies primarily on a profuse variety of self reports and literary and descriptive terms that suggest meaning rather than define it, and because those reports are not deconstructed into underlying facts of experience as reflected in somatic, perceptual, and neurological events. Thus, explanations of flow cannot be generalized to such related motivational processes such as meditation, hypnosis, emotion etc., and it links up with no other subject matters in psychology such as learning theory, neurobiology, etc.
Although the concept of flow ultimately fails as a scientific term, this however is due not to any lack of validity in the self reports which clearly demonstrate that a very rewarding emotional state does occur during certain perceptual circumstances. Because the interpretation of the flow response traces almost exclusively to Csizkszentmihalyi’s extensive writings on the subject, its failure stems from his decidedly literary explanatory style that poses as rather than embodies the principles of science. More pointedly, because flow represents a mere taxonomy of real and obscure events rather than a bona-fide scientific theory, the empirical facts surrounding flow have been rendered unapproachable by the different types of inquiry which could best explain it. That is, since the behavioral or informative events that parallel flow have been ignored or only partially described, this restricts experimental inquiry to methodologies that cannot adequately explain flow, and indeed seem to confirm the description of flow as a mere collection of separate motivational processes. These methodologies are evidenced in Csikszentmihalyi’s own studies on flow.
In addition to utilizing self reports to validate the flow experience, Csikszentmihalyi compared groups of self reports from individuals in different work and social environments. Using an experience sampling method (ESM) (Csikzsentmihalyi 1990) consisting of self reports of individuals that were polled as they engage in daily activities, optimal experience or flow was found to correlate with certain structures of life experience. This between group experimental design distinguished differences between classes of events, such as the general circumstances that were likely to produce flow, but it was nonetheless incapable of delineating the processes that constitute those events. Since flow has been defined as a steady state behavior, it is by definition a transition-less state, and resists decomposition into constituent parts. As we have noted, the impracticality of decomposing such behavior may also reflect the lack of a higher order descriptive language. For example, Aristotle came to the conclusion that heavier objects fell faster than lighter objects by simply observing how classes of heavy objects fall faster than similar groups of lighter objects. His de facto between group design could not account for the behavior of individual objects in time and their behavior when subjected to varying independent variables such as wind resistance, force, acceleration, etc. He was also incapable of employing a within group analysis of how individual objects behave because he had no mathematical language to describe force, acceleration, and inertia. A comparison of how separate groups of individuals respond to different perceptual conditions can only provide general conclusions as to what conditions they will likely respond to, with specific behavioral processes remaining undetermined. This use of separate groups in between subject designs "destroys the continuity of cause and effect that characterizes an irreversible behavioral process…." (Sidman, 1960), and "destroys, confounds, or omits the significant data of moment to moment rate changes in a single organism’s behavior" (Krantz, 1971). A key to a within group trial is the reversibility of a behavioral phenomenon. Reversibility means that a steady state behavior (flow) may be mapped to specific informative patterns or contingencies, and can be recovered over and over again after the subject has been shifted temporarily to other conditions and then returned to those original conditions.
The flow response has never been subjected to a within group analysis because dynamic informative patterns have been generally ignored in the flow literature. The flow experience is elicited by the perceived matching of demand and skill, yet in all theoretical models for flow, variations in this demand/skill equivalence are not mapped in time, nor are they mapped to the differing objective consequences of those real and anticipated variations. Indeed, the individual’s present and anticipated perception of these consequences represents a major independent variable that correlates with significant aspects of the flow response. For example, a perceived matching of demand and skill may occur among rock climbers (Csikszentmihalyi, 1977) and internet users (Novak and Hoffman, 1997), yet the anticipated consequences of a perceived overmatching of demand to skill would be very severe for a rock climber (namely, falling off the mountain), and insignificant for the web surfer. An accounting for the projection or anticipation of the possible results of behavior in time transforms the static demand/skill match into its dynamic representation as a behavioral contingency. That is, climbing a mountain or surfing the net not only gains anticipated results, but performing these skills at a high rate that is characterized by strict attentiveness to these tasks also avoids the consequences that would occur if those behaviors were less attentively performed. In addition, the fact that a climber and surfer also anticipates future correlations between demand and skill, and that these correlations will implicitly vary, suggests that it is not the matching of demand and skill that elicits flow, but the slight anticipated mismatching of demand and skill that will occur in the near future. Finally, the emotional concomitant to this behavior, namely somatic responses such as muscular tension and relaxation may also result in dynamic changes in contingencies (or environment-behavior outcomes), and as we will later note, may be interpreted statically as somatic markers, or dynamically as operant behaviors.
Only a within group design could address why flow is often reported to occur prior to or in anticipation of entering a flow channel, or how flow declines or extinguishes subsequently to performing within that channel. Ultimately, the between group methodology utilized to explore flow reflects the fact that since flow is defined as a steady or transition-less state, it can only be defined by the methodology which in effect confirms that presumption. The procedure acts to sustain a self-fulfilling prophecy, since any results provided by such a procedure cannot disconfirm the multi-process explanation for flow. Ultimately, the between group design used by Csikszentmihalyi is incapable of determining how flow is acquired, how it is sustained, and how it declines. Moreover, self reports are also inadequate indices of flow, since they are not true empirical data, but are interpretations of empirical events that do not become any more definable when averaged. Moreover, these events are understood through their metaphorical representation in similar experiences. Thus we can understand a sense of playfulness, control, and ecstasy since they are analogous to similar experiences that have occurred in the same or different circumstances. Ultimately, the content of flow, namely play, ecstasy, attention, etc. is derived from self reports, and constitute individual processes that are "built into" the perceptual parameters of demand and skill, and are inferred rather than empirical constructs. Unfortunately, this implicit conclusion of a content specific character of the ‘flow channel’ is not supported by modern cognitive science. Lewicki (1998) summarizes this position thusly: "In hardly any experimental procedure do cognitive researchers assume that they can directly learn how humans process information by simply ‘asking’ them to report the contents of the procedural knowledge they follow. No matter how cooperative and well trained our subjects are, they cannot tell us how they go about processing information (e.g. how they encode shapes of objects in three-dimensional space, or how they generate esthetic judgments). This is because subjects not only do not know how they do all those things but have never known it, and they do not have the slightest idea of how they learned all those information-processing algorithms and heuristics that are involved in the cognitive "software" that is indispensable for their psychological functioning. At the same time, there is no doubt that most of this procedural knowledge and skills result from experience and thus that they have been learned at some point." The stimulus presentations in any experiential situation yield much more learning than can be subsequently reported verbally by subject (Greenwald, 1992). Thus, the ESM procedure can only yield partial information at best regarding the content and causes of flow. Ultimately, the effectiveness of the ESM procedure has been greatly overstated since the content of self reports has been incorrectly assumed to be representative of separate processes, with little or no interpretation of the conscious and non-conscious information that is mediated by neural (attention), perceptual (demand/skill match), and somatic (relaxation) events. Because informative variables are barely considered in all flow studies performed utilizing the ESM procedure, the theoretical conclusions of such studies are nearly worthless.
By ignoring the informative variables that are correlated with learning processes, Csikszentmihalyi simply could not derive from these reports any higher order language to thoroughly describe the true empirical correlates to flow, namely the neurological, somatic, and behavioral events that comprise it. The fact that he scarcely defined these events is attributable in large measure to the inadequacy of his method of exploring the flow experience, and not to the unavailability of a better language or the inexplicability of flow. To say that water is wet does not describe water, as an allusion to processes does not describe a process. The subjective language used by the individuals who report flow, even when averaged, is self contradictory, complex, and confusing. However, if the independent variables which correlate with flow are defined abstractly as patterns of information, then flow may be mapped to that information, and easily explained. The emphasis on how behavior dynamically changes with information, in contrast to how it is statically represented by information, represents the guiding principle of behavioristic psychology. Behaviorism’s unique emphasis on information and behavior shifts the emphasis from a study of static steady state behavior to how behavior varies with all values of the independent variable (i.e. information), and how it can be recovered when those values are repeated. As compared to the between group designs which are prevalent in experimental and social psychology, a within group design can trace behavioral processes in time, but is practicable only when dependent and independent variables can be precisely defined and easily observed or reliably inferred. With the availability of higher order descriptive language for behavior, we can now do just that.
A Walk in the Park
The characterization of flow as a holistic experience characterized by attentiveness, playfulness, a unique state of consciousness, and intrinsic satisfaction merely describes a taxonomy of real or imagined processes, and does not allow for the derivation of these processes from any higher order principles. The fact that these processes are therefore implicitly indivisible renders them useless as scientific criteria unless they may be integrated into and derived from a higher descriptive language for behavior which can map all somatic and neurophysiological responses.
If all behavior may ultimately be mapped to information, then flow is no different. Indeed, it may be very simply derived from as ordinary a behavior as walking with a ball in hand. For this mind experiment, let us consider a ball lying in a sports park. The ball of course has color, texture, and shape, and creates a perceptual field through its different points of reference that may be represented by fences, grandstands, or chalk markings on a field. The ball also creates has a variety of possible histories and futures, and this temporal dimension is created through a Bayesian map that can be inferred from the facts derived from the situation or one’s personal history. Thus we can perceive different branching histories for the ball, and trace backwards the ball from how it came to rest in the field to how it was manufactured. Similarly, we can perceive different futures for the ball that would project who would likely pick it up and how it could be used in the future. The myriad number of these perceptual ‘facts’ do not enter consciousness, but rather represent latent behavior-environment relationships or expectancies that come into awareness when they are perceived as instrumental in achieving important or salient outcomes. In this case, let us assume that the individual is promised a small loving cup and monetary reward by simply picking up the ball and walking to a goal a few hundred feet away. Walking to the goal achieves the money and a little cup, and behavior is presumably motivated by this extrinsic reward. Because the behavior is otherwise well practiced and unremarkable, he does not consciously appraise the path he is taking, and he can think of other things besides his walk to the goal. Indeed, the task would otherwise be boring or aversive in lieu of any need to mentally rehearse his behavioral options or the implications of the goal event. Now let us complicate the individual’s task by forcing him to avoid or dodge other individuals who are running towards him with the intent of taking the ball away. Now, the ball carrier must simultaneously consider the many running patterns that will allow him to avoid his pursuers. These multiple patterns of information, although otherwise always available or latent, are now salient to him and are actively considered. This increase in the amount of information considered requires a high degree of mental alertness in order to infer optimal choices from the myriad options available. This active survey or appraisal of alternative choices represents the processing of a great number of environment-behavior relationships in time, and thus represents a highly reinforcing situation that is felt as a state of elation or pleasure. The situational requirement of processing multiple informative precepts in a task environment is called a ‘game’, and because this information is mediated by behavior rather than a discrete physical object, this process of cognitive selection becomes intrinsically desirable or rewarding. If the individual must choose his running pattern within some time constraint, he must therefore process more information in time, and the behavior will become more narrowly focused and more reinforcing as that time constraint narrows. If this time constraint is too narrow, the demands of the situation will be perceived to overmatch his perceived skills, active cognitive appraisal will become less useful, and muscular tension will occur. This tension or anxiety will promote the quick and crude decision making that will allow the individual to escape the situation or make decisions more rapidly (Williams et al. 1988). In other words, anxiety may occur and detract value from his content of the current precepts he is appraising. If some of those precepts were extraneous to his appraisal of the proper running moves, such as a momentary reverie about a personal relationship, then anxiety will increase his cognitive effectiveness by eliminating extraneous thoughts and sharpening his focus on the precepts relating to running patterns. If however no extraneous thoughts existed, then anxiety would reinforce choices that are less optimal, and the effectiveness of his behavior will decline. If the present and anticipated demands of the situation briefly and slightly under or over matched his perceived skills, this pattern of small and momentary variances may otherwise signal momentary inattentiveness or anxiety, either of which will be detrimental to performance. It is hypothesized that in such a situation relaxation would be elicited instead, and act as a somatic marker to keep the individual on his course of action rather than diverting him from it, and would be incompatible with or countercondition any anxiety elicited by intermittent or anticipated variances in the perceived matching of demand and skill. Finally, the intensity of this relaxation response will be scalable with the perceived importance of the reinforcing event. Thus it would be predicted that the ball carrier will find the experience more pleasurable if the stakes for winning were much higher. It is this combination of a highly reinforcing state and a varying intensity of the somatic marker of relaxation that is interpreted as a flow response.
Flow represents the performance of a highly reinforcing behavior that is defined and sustained through a sensitivity to its perceptual content and the pleasant somatic (relaxation) and neurological states that accentuate the importance of that content. That is, flow represents the combined feeling of relaxation, the neurological activity that parallels attentive alertness, and the near simultaneous processing of multiple chains of similar information. Moreover, the separate motivating processes or constructs that have been hypothesized to constitute flow may be derived from the combinatorial aspects of the perceptual, behavioral, and neurological events that comprise flow, and prove to be mere subsets of empirical events. Thus play represents merely the rapidly successive perception of salient patterns of information, and attention, arousal, or involvement represents the homogeneity of these patterns plus alertness. A sense of control denotes the ability to correctly choose fitting patterns of information, and the ‘autotelic’ (Csikzsentmihalyi 1991) or self reinforcing aspect of intrinsically motivating flow states is due to abstract rather than discrete perceptual markers. Skill represents the class of responses an individual may call upon to achieve a goal, and challenge represents the consequences that may occur if those responses are not performed in a certain form and timeliness. Finally, the static matching of skill and challenge can be dynamically represented as informative dependencies or contingencies, and the special state of consciousness in flow represents the repeatedly reconstructed biological and perceptual state that comprises these events (Damasio, 1994).
Flow and Behavior Analysis
The description of a behavioral event is coextensive with a description of the cognitive state that parallels those events, and an empirical as opposed to common sense analysis merely provides a more thorough description of those events that represents their dynamic and temporal as opposed to static and timeless qualities. Thus a baseball in mid flight causes the movements of the outfielder who will catch it, yet an empirical explanation of an outfielder’s behavior does not attribute that behavior to a reaction to the stimulus event of a ball momentarily frozen in space, but rather to his unconscious appraisal of its momentum, direction, and speed. Similarly, the mere sight of a lion does not cause a sudden urge to run away, but rather is dependent upon an appraisal of the lion’s intent, distance from us, and direction of movement, and our own estimate of whether running is the most effective means to make our escape. Moreover, the fact that other concurrent neural, somatic and visceral responses (e.g. attentional arousal, muscular tension, hormonal reactions, etc.) are also initiated and sustained by similar cognitive appraisals as well as mediate cognitions demonstrates that seemingly simple behavioral phenomena may only be understood by decomposing them into the respective informative events that are mediated by the mind and the body proper. The fact that all behavioral and emotional events can be decomposed into patterns of information mediated by the varying inputs of perceptual, musculoskeletel, visceral, or neurological events renders the explanatory power of undifferentiated global definitions of such events of marginal utility. The concept of ‘stress’ for example represents a cognitive appraisal of a situation that parallels various permutations and degrees of muscular, hormonal, and neurological reactions. Thus a stressful event, if undefined, may have a score of very different meanings, and thus mean little or nothing at all.
To illustrate, stress may represent a challenging or demanding event that evokes no concomitant physiological response, save for a heightened attentiveness, or that appraisal may seem automatic or reflexive, as in a ‘flight or fight’ response, and elicit hormonal, visceral, and muscular reactions. Finally, that appraisal may be deliberative or conscious, and initially elicit mainly muscular reactions, as when a student becomes tense in anticipation of a final exam. Although information is at the root of the many different responses that may be termed stress, that information may nonetheless be mediated by different neural structures that prime an individual to rapid and more deliberative responses. Thus the fear or startle reactions that comprise the flight or fight response occur through the intermediation of mid brain structures such as the amygdala that are sensitive to gross aspects of a stimulus, such as novelty, shape, and movement (Ledoux, 1996), whereas the muscular tension characteristic of the workaday demands of life occur with relatively less rapidity, and are elicited by abstract perceptual events that are assembled by higher cortical brain structures, and may be mapped to environmental contingencies.
The elimination of false process distinctions between intrinsic and extrinsic motivation and the innumerable subprocesses such as flow, stress, play, etc. which comprise them moves an analysis of behavior from mere compilations of poorly defined processes to information itself. The use of information as the common denominator for overt and covert behavior allows for the better understanding of behavioral situations and outcomes, and provides for better and more parsimonious procedures for the control of behavior. The practical implications herald a potential recasting of the Babel of conflicting paradigms that represent the subject matters of social psychology, humanistic psychology, psychotherapy, etc. into a far more parsimonious model that embodies a common language for behavior that all psychologists may share. The ability to decompose behavior into various physical modalities that mediate information ultimately forces psychologists to talk to rather than past each other, and ultimately resolves perennial debates through the precise definition of terms. As importantly, the move to informative rather than process distinctions permits testable hypotheses and practical procedures, and allows for the decisive refutation of hypotheses. The ability of a theory to be decisively falsified (Popper, 1959) distinguishes good theories from bad, and allows for the resolution of many theoretical problems that are otherwise impossible to resolve. Thus we have noted that theories explaining the flow response are essentially unfalsifiable because they are barren of significant testable conclusions, since they rest on the ever shifting terminology and meaning of self reports.
In particular, an information processing or expectancy theory for the flow experience provides testable procedures that permit the derivation of new and powerful procedures for self control. Besides the theoretical description of flow, its practical implications are the useful procedures which may be derived from the flow experience. Crucial to the construction of such procedures is an analysis of how flow may occur independently of the perceptual setting (i.e., demand/skill match) which elicits it. As has been noted, no studies have been performed to explore how flow declines after its sustaining perceptual parameters have ceased. Logically, it would be predicted that the emotional components (relaxation, arousal) of flow would abruptly end. But evidence from self reports and neuro-psychology demonstrates this is very likely not the case.
For example, for individuals who experience similar meditative states, relaxation does not cease, but often continues for quite some time. This emotional trait effect (Goleman and Schwartz, 1976) describes the continuing presence of relaxation subsequently to the cessation of the circumstances that originally created that relaxation. The trait of relaxation refers to the continuation of a relaxed state without conscious awareness or intervention, or in other words sustained by nonconscious processes. However, the nonconscious processes that sustain relaxation, if ultimately referred to information, must reflect the recreation of the proprioceptive stimuli that constitute the feeling of relaxation, and the recreation of the perceptual set which elicited relaxation. That is, relaxation may be evoked through the nonconscious elicitation of relaxation due to one’s being sensitized towards the somatic events that comprise it, or by recreating the stimulus events or patterns of events that elicit and sustain relaxation. These events constitute the emotional memory of an event, and will decline in their evocative potential in time as an inverse function of the duration, intensity, and value of the emotional event. Thus the anxiety attending a traumatic event will likely be sustained far longer subsequently to that event than a mildly stressful event.
Finally, the greater number of stimuli that are associated with a response, the more likely that any given environment will contain some of those stimuli, and hence the response will reoccur and/or persist. This ‘over-expectation’ effect, or behavioral momentum (Nevin, 1992) would assign a discriminative function to otherwise neutral stimuli that have been associated with the response. Thus, the continuation of an emotional response long after its proximal causes have ceased may be attributed to remaining in the original environmental setting (office, laboratory) of that response.
This analysis becomes a bit more complicated when situations that elicit the sustained release of dopamine are considered. The positive affect caused by unexpected rewards has been attributed to the release of the neuromodulator dopamine, yet dopamine release continues long after dopamine cells have stopped firing (Ashby, Isen, and Turken, 1999). Although dopamine release has been noted to occur up to thirty minutes after the stimulation of dopaminergic systems, it remains unclear how emotional memory or behavioral momentum may facilitate or inhibit the degree and persistence of the release of dopamine over time.
It is important to understand that the evocative power of memory is not due to the decay of a memory trace, but to incompatible information available which elicits incompatible response, or acts as a signal for no response. Thus, an individual will more likely continue to be anxious if he remains in the physical context of that anxiety (e.g. the scene of a crime, accident, etc.) than if he was removed to a neutral location that signaled no danger. Indeed, if the individual was confronted with some cue that reminded him of the original event, such as a memento of a romantic evening, the emotional memory (including of course the emotion it signifies) of that evening may also be recalled. Finally, these memories may be rapidly recovered when an individual anticipates the imminent recreation of the emotional event. These ‘as-if’ emotional memories (Damasio, 1994) represent a preparatory response that enables an individual to rehearse all components, including emotional components, of a response prior to revisiting the context which will directly elicit that response. Furthermore, the rehearsal of these stimulus compounds, which includes emotion, is equivalent to a reinforcing event (Davis and Doogan, 1988), and provides survival value by fixing in memory important patterns of information, and preparing the individual for imminent action.
The Theory and Practice of Flow
As has been noted, the ‘flow theory’ as elucidated by Csikszentmihalyi is not a true scientific theory, since it contains severe difficulties in its logical syntax and semantics. In addition, because Csikszentmihalyi imports meaning through his ample use of metaphor, he succeeds in capturing only the phenomenological reality of the experience. Just as one may in turn explain ‘green’ as a property that inheres in things in the world and on the neural level a multiplace interactional property, the phenomenological and neural levels of understanding provide two different modes of understanding. The first is in terms of everyday experience and the second is in scientific terms (Lakoff, 1999). The confusion occurs when the flow theory is assumed to represent a science, when it does not.
As conceptualized by Csikszentmihalyi, the flow theory does not suggest any testable hypotheses that can verify what flow is or how it can be manipulated. However, if a significant dependent measure of flow is alertness and relaxation, and if both can be mapped to dynamic patterns of information that correspond to a demand/skill match and may be grounded to actual neural events, then a theoretical model for flow can be derived, along with valuable procedures for behavioral control. As is generally accepted, a feeling of ecstasy, calm, pleasure, etc. is often elicited when an individual performs some important task that continually tests him to the limit of his capabilities. This demand/skill match represents merely a behavioral contingency that engages one’s entire perceptual resources, and is otherwise represented by a high volatility in perceptual discrimination since the individual must constantly shift between multiple salient precepts. This high level of involvement finds a neural correlate in a high level of information processing which is felt as a high alertness or a state of elation. The more salient the task, the greater will be the number of attributes of that task that must be attended to, and hence the more accentuated will be the activity of midbrain dopamine systems. Ironically, Csikszentmihalyi was correct with his identification of flow with focused attention. His error was in defining attention in literary metaphor rather than grounding it in actual neural processes. The flow response is a discrete event because it’s metaphors are discrete (psychic energy, order in consciousness). But as defined as an actual neural process, flow becomes an indiscrete event, and becomes more pleasurable or intense as the relative salience and number of cognitive precepts increase in time. Thus, attention given to moderately unsalient perceptual events (e.g. driving a car) will be less pleasurable than events that have high salience (e.g. playing an ‘addictive video’ game, watching an engrossing movie). The latter in turn will be less pleasurable than events that have an extraordinarily high salience (mountain climbing, performing surgery). The scalability of this pleasurable effect is due to the increase in activation of dopamine mid brain systems. The more activation of these systems, the more intense the pleasure.
Secondly, along with the pleasurable sensation accompanying focussed alertness, it is hypothesized that relaxation also will occur, and will also scale with the level of salience of attended perceptual events. As the presence of dopamine moderates alertness and through its hedonic quality signifies value, relaxation responses perform much the same function. Relaxation, as a somatic marker, will also signify the value of the cognitive precepts is parallels, help fix attention on those precepts, and of course will be felt as a pleasurable emotion. The degree of relaxation will also scale with the importance of the task.
Learning Theory and Philosophy
Flow does not represent a special state, but is rather a category reserved for a particular aggregation of unremarkable perceptual, behavioral, and neurological events. However, the elements that are mundane when considered separately combine to create an emergent or holistic experience that is optimally rewarding and unique. As with the stuff of life itself, the components of flow are ultimately nothing special or mysterious. Indeed, the very aura of mystery that has long surrounded the components of flow has obscured the practical and philosophical implications of the flow experience.
The psychology of flow and the optimal experience it embodies represent in reality the search for the factors in life that combine to reinforce optimally, and these factors ultimately derive from elemental sensitivities that derive from our evolutionary past. If behavior is best defined through the abstract language of information as guided by nativistic sensitivities, then our values may be defined as abstractly, and value becomes denoted not in objects, but in information. The implications of flow range in a continuum from the practical procedures and innate preferences that guide the construction of individual behavior to the formation of social groups and the design of cultures. But to illustrate these implications, let us return to our mind experiment.
When we left our ball carrier, his effort to reach the other end of the field was motivated by a monetary reward and a loving cup. However, the relatively insignificant aspect of a loving cup in hand gains greater currency if there was a stadium audience that voiced approval if the ball carrier succeeded in his attempt to reach his goal. The cup signifies the expectancy that would reflect his indirect control over the behavior of others, and his rehearsal of this precept and its many offshoots would therefore be very reinforcing. Moreover, after the game has finished, the loving cup acts a perceptual marker to elicit the memories of his success, and continues to reinforce, even though the game occurred in the past. The game is very reinforcing because of information that is created virtually, and this information may be suggested by memory and through the memories elicited by the presence of a physical object. The non-material basis of reinforcement implies that reinforcers must be judged in terms of how they mediate information, and not through their physical basis. This indiscrete aspect of reinforcing events is further extended by the fact that perceptual markers for information may continue to elicit memories of past information, and this adds to their ultimate value. Thus, the ball carrier would find it more reinforcing to win his prize in an arena of real people that provides recurring informative value than to repeat the same experience with a virtual audience (such as in a video game) that has the mere semblance of meaning.
The implication of modern learning theory is that behavior is dependent upon the perception of patterns of information that can be reduced to a perceptual geometry, and that the type of information that is selected is chosen due to a biologic sensitivity towards stimulus patterns and events that have a rudimentary survival value. This value in turn is ultimately chosen by natural selection. Events are valued in proportion to the type and amount of information they mediate in time, yet the physical properties of those events serve as mere markers for value rather than as intrinsic possessors of value. That is, value can be denominated in checkers and stamp collecting just as well as it can be measured in a dictatorial power over people.
The native ability to perceive causal relationships that reflect personal control has been accompanied by a tendency to select such information because of its intrinsic survival value. Unlike the specific sensitivities towards food, drink, and sex that translate into behaviors that provide individual benefit, rudimentary sensitivities have general and not specific survival value, and may weigh against the survival of individuals, but yet be selected because of the value it provides to the group. Hence whole armies can march to their deaths for God and country, yet their demise has personal meaning and value that is denoted in the appraisal of other people who continue to survive. Thus individuals who have highly developed ‘senses’ of honor, justice, courage, etc. simply have the ability to model the minds of other people and how they regard or would regard one’s behavior. This empathic quality reflects the ability to model the intentions of other people and the emotions that parallel them. For example, an athlete who is encouraged by a roaring crowd not only understands what is in their minds but also can model or reflect their feelings. His awareness of his surroundings reaches a culmination in his knowledge of and a resulting feeling for others. Therefore, an empathic individual not only has a greater insight into the interpersonal implications of his actions, but he also processes more information and is more reinforced than an individual who is less or non-empathic. Empathy improves decision making because it provides reliable information as to the cognitive states of others, but also our resulting behavior is determined or reinforced by an estimate of those states. Moreover, this estimation does not and often will not result in any change in the behavior of other people, and behavior can be maintained through the virtual rather than actual reflection of the behavior of others.
Reinforcers are specific to the quality of the information rather than the aspect of the medium that signifies or transmits that information, and are also specific to our capability to model causality in terms of future events and in terms of the future behavior of others. In time, the more positive information that can be mediated by the greater number of events, whether those events represent physical objects or the mind-sets of other people, the more reinforcing or rewarding will one’s life be. Thus the meaning in life represents the continuous creation of value that is sustained through our sensitivities to specific types of information and by our ability to model the present and future outcomes in our physical world and in the mental worlds of other people.
If reinforcers are information specific, then the progress and value of cultures are judged by the type and amount of information that is mediated by its artistic and social institutions. Complex cultural artifacts such as classical music and literature, representational art, classic and folk dance, complex games (e.g. chess, football) etc. mediate more information than simpler forms of music, art, dance, and sport, and are therefore potentially more reinforcing. Moreover, the individual possession of knowledge of these genres becomes a medium of exchange between other people. Thus, the more people have the capability and ease to communicate common interests, and in particular about complex things, the more they will be reinforcing to each other. Finally, the advancement through cultural design of multiple interests shared in common increases social interaction between all people, and the resulting ability to empathize with more people eliminates social and class distinctions.
Descartes Error and the Tragedy of Modern Psychology
The postulation of separate motivating processes implies that such processes are incommensurate, and cannot be translated into one another or be practically derived from higher order principles. If motivating processes can only refer to a literal bundle of nerves rather than to a unitary set of principles, then the impracticality of referring separate motivational processes to separate neural structures condemns psychology to forever define mental processes in terms of an ever growing and convoluted list of discrete metaphorical motivational states, since it cannot derive them from a set of principles that represent the elementary conscious and non-conscious information that is dynamically mediated by somatic, perceptual, and neurological events. In social and humanistic psychology, the persistent refusal to systematically account for the rudimentary patterns of information that underlie behavior has resulted in a massive corpus of confounding, confusing, and redundant research and theory that is described not in precise logical terms, but in the tonnage of an endless train of journalistic jargon.
The philosophical lineage of this multi-process view of human nature derives from Rene Descartes, who rejected a monistic view of nature in favor of dualistic split of human nature that posed a separate seat of reason and of emotion. However, as a principle of science, this dualism was inherently unscientific because the separate terms were unconvertible and irreducible. In the physical sciences the postulation of separate forces are recognized as mere temporary waypoints until higher order principles are discovered, and its progress has always been measured by the derivation of ever simpler governing principles. Cartesian philosophy effectively put arbitrary limits to this process, and preserved a dualism that has assumed the modern guise of an ever increasing list of insubstantial motivating forces that has made much of modern psychology unreadable, incomprehensible, and useless. In contrast to the obfuscating principles that have proven devastating to the science of psychology, higher order principles may now be advanced that can describe all of the facts of behavior. This monistic viewpoint assumes that all of nature, including human nature, derives from one substance, and may be described by mathematical law. It is built upon an empirical methodology that describes reliably recorded facts, and does not jump to abstractions or universals. Through experiment and observation, it recognizes that all human behavior, including emotion, may be described through a perceptual geometry, and may be derived from first principles. Finally, it recognizes the native values that our biologic heritage impels us to pursue, and that the purpose of life is to preserve value.
Remarkably, these principles are not new, and were advanced by a contemporary and student of Descartes who deduced a perceptual geometry for behavior over two hundred years before the birth of experimental psychology. His recognition that biologic value is essential to human life led him to believe that any philosophy must lead to the embrace of melioristic principles. To this end Baruch de Spinoza developed a philosophy built upon a rigorously deductive logic that explained the facts of his experience and the purpose that experience serves, and named it ‘The Ethics.’ The description of behavior begins in physiology, advances to psychology, and ends in philosophy. The unification of all knowledge hinges ultimately on what is a reinforcing event, or what is value. From that knowledge, all psychology may be derived.
Ashby, F. G., Isen, A. M., and Turken U. (1999) A Neuropsychological Theory of Positive Affect and Its Influence on Cognition, Psychological Review, 106, (3), 529-550
Barber, T.E. (1969) Hypnosis: A Scientific Approach, New York: Van Nostrand
Birch, L.L., D. W. Marlin, and J. Rotter (1984) Eating as the ‘means’ activity in a contingency: effects on young children’s food preference. Child Development 55(2), 431-439
Bolles, R. C. (1972) Reinforcement, Expectancy, and Learning. Psychological Review, 79, (5), 394-409
Bolles, Robert C. (1976) Theory of Motivation. 2nd ed. New York: Harper and Row
Brown, D.P. (1977) A model for the levels of concentrated meditation. International Journal of Clinical and Experimental Hypnosis, 25: 236-273
Butler, R., and M. Nisan (1986) Effects of no feedback, task-related comments and grades on intrinsic motivation and performance. Journal of Educational Psychology 78(3), 210-216
Canter, Rivers, and Storrs (1985) Characterizing user navigation through complex data structures, Behaviour and Information Technology, 4(2) 93-102
Csikszentmihalyi, M. (1977) Beyond Boredom and Anxiety, Jossey Bass
Csikszentmihalyi, M. (1978) Attention and the holistic approach to behavior. In Pope and Singer (Eds.), The Stream of Consciousness. New York: Plenum
Csikszentmihalyi, M. (1990) Flow, The Psychology of Optimal Experience. Harper Collins
Csikszentmihalyi, M. (1997) Finding Flow, Basic Books
Czikszentmihalyi, M. (1975) Play and intrinsic rewards. Journal of Humanistic Psychology, 17(3): 41-63
Damasio, Antonio R. (1994) Descartes Error: Emotion, Reason, and the Human Brain. Avon
Davis, R. T. and Dougan, J. T. (1988) The phylogeny of information processing. In Robert C. Bolles and Michael D. Beecher (Eds.) Evolution and Learning, Lawrence Erlbaum Associates-London.
Deci, E. L. and Ryan, R. M. (1985) Intrinsic Motivation and Self-Determination in Human Behavior. New York: Plenum
Deci E. and Porac J. (1978) Cognitive evaluation theory in the study of human motivation, in The Hidden Costs of Reward (edited by M. Lepper and D. Green), Erlbaum, New York
Deikman, A. J. (1963) Experimental meditation. Journal of Nervous and Mental Disturbance, 136: 329-343
Dennett, Daniel (1987) The Intentional Stance MIT Press
De Sousa, Ronald (1987) The Rationality of Emotion. MIT Press
Donahoe, J.W. and D. C. Palmer (1993) Learning and Complex Behavior, Allyn and Bacon
Durstewitz, Daniel, Marian Kelc, and Onur Gunturkun (1999) A neurocomputational theory of the dopaminergic modulation of working memory functions, Journal of Neuroscience, 19(7): 2807-2822
Fenwick, P.B. et al. (1977) Metabolic and EEG changes during transcendental meditation: an explanation. Biological Psychology, 5(2): 101-118
Ghani, J. and S. Deshpande (1994) Task characteristics and the experience of optimal flow in human-computer interaction, The Journal of Psychology, 128(4), 381-391
Goleman, Daniel J. and Schwartz, Gary E. (1976) Meditation as an intervention in stress reactivity. Journal of Consulting and Clinical Psychology. 44(3): 456-466
Goleman, D. J. (1995) Emotional Intelligence, Bantam
Graef, R. Csikszentmihalyi, M., & Gianinno S. (1983) Measuring intrinsic motivation in everyday life, Leisure Studies, 2, 155-168
Greenwald, A.G. (1992) Unconscious Cognition Reclaimed, American Psychologist, 766-775
Grolnick, W. S., and R. M. Ryan (1987) Autonomy in children’s learning: an experimental and individual difference investigation. Journal of Personality and Social Psychology, 52: 890-898
Harris, Marvin (1979) Cultural Materialism, Random House
Hearst, Eliot (1975) The classical-instrumental distinction: reflexes, voluntary behavior, and categories of associative learning. In W.K. Estes (Ed.) Handbook of Learning and Cognitive Processes, Vol. 2, New York: Wiley and Sons
Hilgard, Ernest and Bower, Gordon H. (1975) Theories of Learning. New York: Appleton Century-Crofts
Hillner, Kenneth P. (1979) Conditioning in Contemporary Perspective. New York, Springer
Holmes, David S. (1984) Meditation and somatic arousal reduction. A review of the experimental evidence. American Psychologist, 39(1): 1-10
Holmes, David S. (1988) The influence of meditation versus rest on physiological arousal: a second evaluation. In Michael A. West (Ed.) The Psychology of Meditation, Clarendon Press-Oxford
Horvitz, J. C., Stewart, T., and Jacobs, B. L. (1997) Burst activity of ventral segmental dopamine neurons as elicited by sensory stimuli. Brain Research, 759, 251-258
Jeannerod, M. (1994) The representing brain: Neural correlates of motor intention and imagery. Behavioral and Brain Sciences, 17(2): 187-245
Kimble, G. A. & Perlmuter, L. C. (1970) The problem of volition. Psychological Review, 77, 361-384
Kirby, L. D. and Smith, C. A. (1997) The structure of appraisal. Twentieth Annual Conference of the Society of Southeastern Social Psychologists. Research Triangle Park, NC.
Koepp, M.J., Gunn, R.N., Lawrence, A.D., Cunningham, V.J. Dagher, A. Jones, T., Brooks, D.J. Bench C. J., Grasby, P.M. (1998) Evidence for stiatal dopamine release during a video game. Nature, 393: 266-268
Kohn, A. (1993) Punished by rewards: the trouble with gold stars, incentive plans, A’s, praise, and other bribes. Boston: Houghton Mifflin
Kosslyn, S. M., Pinker, S. Smith, G.E. & Schwartz, S. P. (1979) On the demystification of mental imagery, Behavioral and Brain Sciences, 2: 535-582
Krantz, D. L. (1971) The separate worlds of operant and non-operant psychology. Journal of Applied Behavior Analysis, 4, 61-70
Kruglanski, A.W. (1978) Endogenous attribution and intrinsic motivation, in The Hidden Costs of Reward (edited by M. Lepper and D. Greene), Erlbaum, New York.
Kuhn (1970) The Structure of Scientific Revolutions, University of Chicago Press
Lakoff, George, and Johnson, Mark (1999) Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought, Basic Books
LeDoux, Joseph (1996) The Emotional Brain, Simon and Schuster
Lewicki, P., Hill, T., & Czyewska, M. (1992) Nonconscious acquisition of information, American Psychologist, 47, 796-801
Marr, M. J. (1985) ‘Tis the gift to be simple: A retrospective of Mach’s The Science of Mechanics. Journal of the Experimental Analysis of Behavior, 44, 129-138
McGuigan, F.J. (1976) The function of covert oral behavior in linguistic coding and integral information processing. In Kurt Salinger (ed.) Psychology in progress: an interim report. Annals of the New York Academy of Sciences, 270: 57-89
McGuigan, F.J. (1978) Cognitive Psychophysiology. Principles of Covert Behavior. Englewood Cliffs, N. J.: Prentice Hall Inc.
Michaels, P.R. et al. (1976) Evaluation of transcendental meditation as a method of reducing stress, Science, 192(4245): 1242-1244
Miller, N. E. (1959) Liberalization of Basic S-R Concepts: Extensions to Conflict Behavior, Motivation and Social Learning. In S. Koch, ed. Psychology: A Study of a Science, (2), 196-292 McGraw Hill
Miller, N. E. & M. David Egger (1962) When is a reward reinforcing? An experimental study of the information hypothesis. Journal of Comparative and Physiological Psychology, 56, 132-137
Mineka, S. and Sutton, S. K. (1992) Cognitive Biases and the Emotional Disorders. Psychological Science, 3(1), 65-69
Mowrer, O. H. (1960) Learning Theory and Behavior. New York: Wiley
Nargeot, R, D. A. Baxter, G.W. Patterson, and J. H. Byrne (1999) Dopaminergic synapses mediate neuronal changes in an analogue of operant conditioning, Journal of Neurophysiology, 81,4, 1083-1987
Nevin, J. A. (1992) An integrative model for the study of behavioral momentum. Journal of the Experimental Analysis of Behavior, 57, 301-316
Novak, Thomas P. and Donna Hoffman (1997) Measuring the Flow Experience among web users, working paper, Vanderbilt University
Nugent, K. and Mineka, S. (1994) The effect of high and low trait anxiety on implicit and explicit memory tasks. Cognition and Emotion, 8(2). 147-163
Pagano, Robert R. and Warrenburg, Stephen (1983) Meditation, in search of a unique effect. In R. J. Davidson et al. eds. Consciousness and Self Regulation, 3, Plenum
Popper, Karl (1959) The Logic of Scientific Discovery. New York: Basic Books
Port, Robert F. and Van Gelder, Timothy (Eds.) (1995) Mind as motion: Explorations in the dynamics of cognition. Cambridge, Ma: MIT Press
Pylyshyn, Z. (1984) Computation and Cognition. MIT Press, Cambridge, MA.
Rescorla, R. A. (1967). Pavlovian conditioning and its proper control procedures. Psychological Review, 74, 71-80
Schacter, S. & Singer, J. E. (1962) Cognitive, social, and physiological determinants of emotional state. Psychological Review, 69, 379-399
Schultz, Wofram (1998) Predictive reward signal of dopamine neurons, Journal of Neurophysiology, 80, 1, 1-27
Searle, J. (1980) Minds, Brains, and Programs, Behavioral and Brain Sciences, 3, 417-458
Shull, Richard L. (1995) Interpreting Cognitive Phenomena: Review of Donahoe and Palmer’s Learning and Complex Behavior, Journal of the Experimental Analysis of Behavior, 63, 347-358
Sidman, M. (1960) Tactics of Scientific Research: Evaluating experimental data in psychology. New York: Basic Books 249-250
Spinoza, B. (1677) The Ethics. Dover
Staddon, J.E.R. Learning as inference. (1988) In Robert C. Bolles and Michael D. Beecher (Eds.) Evolution and Learning, Lawrence Erlbaum Associates-London
Thagard and Barnes (1996) Emotional Decisions. Proceedings of the Eighteenth Annual Conference of the Cognitive Science Society. Erlbaum, 426-429
Treffner, P. J. (1997) Representation and specification in the dynamics of cognition. Contemporary Psychology, 42, 697-699
Wagstaff, Larry and Hamilton, David (1981) Hypnosis, Compliance, and Belief. New York, N.Y.: St. Martins Press
Wagstaff, Larry, and Hamilton, David (1975) American Journal of Hypnosis, 17(3): 190-197
Webster J. and Martocchio J. (1992) Microcomputer playfulness: development of a measure with workplace implications, MIS Quarterly, 16, 201-226
Webster, Trevino, and Ryan (1993) The dimensionality and correlates of flow in human-computer interactions. In Computers in Human Behavior, 9(2), 411-426
Williams, J.M.G., Fraser, N.W., MacLeod, C., and Mathews, A. (1988) Cognitive Psychology and Emotional Disorders. New York, NY: John Wiley.