Search This Blog

Thursday, August 23, 2007

Freud's Signal Anxiety: A Better Explanation for Damasio's Somatic Marker ?

The Somatic Marker: A Bio-behavioral Explanation


The somatic marker hypothesis is one of the most influential explanations of how covert somatic responses guide decision making or choice. Although somatic markers are purported to derive from learning, the somatic marker has never been examined from the sole perspective of learning. It is the purpose of this article to deduce reliable correlations between the bio-behavioral antecedent and consequential events imputed by the primary exemplar of the somatic marker, namely muscular tension. The viability of the somatic marker of tension as a learned or conditioned response is examined, and correlations between tension and behavior can be used to derive testable hypotheses for behavioral control.

key words: Damasio, somatic marker, autonomic arousal, behavioral, mindfulness, tension, IGT


The concept of the somatic marker is one of the most notable and controversial constructs that explain how somatic states enable and change thought, and in a broader sense, the etiology and function of emotion. Postulated by the neurologist Antonio Damasio (1995) , the somatic marker hypothesis states that a state of arousal or a ‘gut feeling’ mediated by the peripheral and central nervous systems often precedes and influences decision making, and that arousal is in turn modulated or conditioned by unspecified learning principles. Inherent in the experience of arousal is that it acts as a conditioned or discriminative stimulus that non verbally connotes the overall goodness or badness of response options and assists reasoning by automatically parsing the response options that will be rationally considered that lead to a specific goal.

A distinct attribute of Damasio’s analysis is that the dependent and independent variables representative of the somatic marker are not behaviorally defined, thus rendering the somatic marker near impossible to study as a product of learning. A behavioral event will be defined as a discrete and observable response that is directly initiated and modulated across trials by correlated changes in equally discrete and observable informative or discriminative events. These correlations and the predictions derived from them constitute the subject matter of a learning theory. This article will demonstrate that the somatic marker, although ostensibly controlled by learning, is incoherently defined as a learnable or behavioral event, and can only be systematically described from the perspective of learning theory if its dependent and independent measures are precisely defined. This analysis is informed by observable bio-behavioral processes, yet derives the somatic marker from an operational perspective. That is, although a learning perspective can explain covert somatic behavior as evidenced by the somatic marker through an integration of neurophysiological with behavioral events, it may also be used to derive a simple pragmatic model that isolates the personally accessible and manipulable events that permit covert behavior to be predicted and controlled. Conforming to the epistemological principles of a radical behaviorism, this perspective is informed by neuro-physiological processes, but is validated primarily by its predictive power. The virtue of this perspective is that it reduces the number of salient factors in its analysis, and produces easily testable predictions that are of major significance in the self control of emotional states.

The Somatic Marker

This somatic marker hypothesis finds primary support from a now classic experiment surnamed the Iowa Gambling Task, or IGT (Damasio,1995). As described by Tomb et al. (2002), “A subject was presented with four decks of cards. After turning over a card, participants either win or lose varying amounts of play money. Unknown to the participants, picking from two of the decks (‘good’ decks) will result in eventual gain, whereas picking from the other two decks (‘bad’ decks) will result in eventual loss. The task ends after the selection of the 100th card, when most normal individuals have picked more cards from the good than the bad decks. After several rounds of picking cards, it was found that ‘anticipatory ’autonomic arousal, as measured indirectly by the skin conductance response (SCR), was significantly higher for bad decks rather than good.” On the primary basis of this experiment, Damasio deduced that arousal acted to non-consciously alert the individual of the bad deck before its ‘badness’ could be rationally determined.

Arousal is the result of previous socialization or learning, and occurs prior to the conscious consideration of response options to alert one to or pre-determine the ‘goodness’ of a particular response set. In other words, autonomic arousal allows one to make a proper choice between response options prior to their rational consideration. The somatic marker (Damasio, 1995) “forces attention on the negative outcome to which a given action may lead, and functions as an automated alarm signal which says: Beware of danger ahead if you choose an option that leads to this outcome. The signal may lead you to reject, immediately, the negative course of action and thus make you chose among other alternatives. The automated signal protects you against future losses, without further ado, and then allows you to choose from among further alternatives. There is still room for using a cost/benefit analysis and proper deductive competence, but only after the automated step drastically reduces the number of options.”

An independent measure implicit in the original IGT experiment but not controlled was the programmed variance in the frequency and value of individual high and low value cards, independent of the overall value of the card deck. This measure was examined in experiments performed by Tomb et al. (2002), Lin et al. (2007), and Chiu et al. (2008), who in comparing variant models of the IGT experiment noted the correlation of perceived changes in the reward or punishment schedule for good and bad decks with subsequent deck preference. By systematically varying the reinforcement schedule as well as the value of individual cards, the authors arranged that prospective moment to moment performance outcomes would result in positive or negative unexpected or counterfactual results. It was found that the pattern of these results lead to the choice of a favored card deck irrespective of whether the overall or expected value of that deck was good or bad. In these experiments, these counterfactual outcomes are predicted but are never certain, and represent surprising changes or ‘discrepancies’. Thus essentially, these results may be interpreted to mean that the ultimate choice of a deck correlates with the expectation of discrepancies in near term task performance whose positive or negative valence are not necessarily coherent with the overall goodness of cumulative responding ( a corollary result recognizable from personal experience would be gambling behavior, where the choice of a slot machine would be in accord to moment to moment chance or ‘luck’ despite the concurrent knowledge that over time and enough pulls, the cumulative result of this behavior will result in a net loss ). In line with these findings, it is proposed that autonomic arousal somatically marks the goodness of future momentary responding rather than overall responding and may also be initiated by perceived discrepancies in the rate or content of reinforcement. This revised function of the somatic marker diverges from Damasio’s interpretation, and will be hypothesized to conform with the notion of autonomic arousal as a behavioral response. However, an accurate appraisal of this hypothesis requires first a concise definition of what the dependent and independent measures of the somatic marker entail.
It must be noted that although primarily justified through the peripheral measures of the skin conductance response (SCR), somatic markers may not involve the periphery at all, but represent “states of bio-biochemical regulation in structures of the brain stem and hypothalamus” (Damasio, 1995). For example, ‘feelings’ of elation, depression, regret etc. certainly ‘mark’ value, but may do so independent of elevated autonomic arousal. In other words, they are psychologically and physiologically distinct from autonomic arousal. Unfortunately, this expands the dependent measures for the somatic marker to include all neurological events and their affective representations that somatically mark the long term value of behavior, and renders the somatic marker near equivalent to the concept of emotion. In this broader interpretation, the somatic marker hypothesis is not specific to a single event and a single cause, but is rather a taxonomy for all affective events originating in the central and peripheral nervous systems that mark long term value. It follows that tension induced autonomic arousal is only a type of somatic marker, although it is its primary exemplar. Since the somatic marker has been justified and criticized through the IGT experiment and the attendant use of the SCR as the dependent measure, for our purposes a learning based examination of the initiating cause of the SCR and autonomic arousal, namely muscular tension, will be used as a suitable proxy for the somatic marker.

The Dependent Measure of Tension

In the IGT experiment, the dependent measure of arousal was indirectly measured through the skin conductance response, or SCR. Autonomic arousal in turn was initiated through processes of socialization or learning. However, Damasio did not provide a systematic explanation of autonomic arousal and its physiological and cognitive antecedents. That is, the dependent measure of autonomic arousal is an ensemble of processes that include muscular tension, accelerated heart rate, increased biochemical activity, etc., yet the somatic marker hypothesis does not impute the systematic relationship of these changes, and it does not describe how those changes may be modulated as behavioral events by information or learning. A standard explanation for autonomic arousal is that it is mediated by the sustained contraction of small low threshold motor units of the striated musculature (Mcguigan,1991), and can be measured directly through EMG (electromyogram) or through indirect measures of autonomic arousal (e.g., skin conductance response or SCR; galvanic skin response or GSR) elicited by tension. The SCR and GSR are used extensively in bio-feedback relaxation therapies to train muscular relaxation (Yucha & Gilbert 2004). Physiologically it has been well demonstrated that muscular tension instigates autonomic arousal (Gellhorn, 1967, 1972, Jacobson, 1970, Malmo, 1975). Through a bi-directional connection between the reticular arousal system and muscle efferents, a dramatic decrease or increase in muscle activity throughout the body can respectively stimulate decreases or increases in sympathetic arousal.

Problematic for the study of tension induced arousal is that the more appropriate measure for sustained or tonic levels of tension is not the SCR, which measures transient or phasic responses, but the SCL, or skin conductance level.(The SCR indexes the net or phasic change in sweat gland activity in response to a stimulus or event (e.g., presentation of a picture or a sound), whereas the SCL indexes the basal or tonic level of sweat gland activity.) Because the SCR has been used near exclusively in replications of decision making behavior such as the IGT , this has resulted in the most salient and common attributes of tension escaping observation. Specifically, tonic levels of muscular tension are commonly produced under continuous alternative contingencies or choices and modulate avoidance behavior. For example, continuous decision making between alternative contingencies (e.g. doing housework or minding a child, working or surfing the internet, etc.) is associated with sustained or tonic levels of tension that is painful and modulates not effective choice, but avoidance. Surnamed the ‘Cinderella Effect’ (Wursted et al. 1991, 1996; Hagg, 1991; Lundberg, 1999), the continuous tensing of low threshold motor units or muscles (or Cinderella fibers) because of this psycho-social ‘demand’ causes them to eventually fail, and thus recruit other groups of muscles more peripheral to the original group, resulting in pain and exhaustion. This result conforms with McEwen’s model of ‘allostatic load’ (1998), which predicts that tension and arousal will be maladaptive when there is an imbalance between activation and rest/recovery. Specifically, continuous tension results in overexposure to stress hormones, high blood pressure, and resulting mental and physical exhaustion.
If tension initiates and is sustained by autonomic arousal, and if arousal cannot occur without tension (McGuigan, 1991), it follows that arousal cannot be directly derived through any other proximal cause. Specifically, this means that arousal cannot be directly modulated by changes in information, and thus cannot be a behavioral event. A similar reasoning occurs for states of physiological activation that sustain other behavioral events such as running. Thus running is a behavioral event because it is directly modulated by changes in information, but the rapid heart beat and other changes that are concomitant with running are not directly modulated by information, and represent instead S-R mechanisms that are physiologically elicited to sustain the behavior itself.

The Independent Measure of Discrepancy

The IGT experiment entailed an observation of a subject’s performance under separate response contingencies, each signified by manipulations (card flipping) of a separate deck of cards. Performance under a primary contingency, namely pulling cards from a single deck, would result in positive and negative discrepant changes (i.e., good cards and bad cards) that are perceived, and through experience, are expected to be perceived. The role of unpredicted positive changes in discriminative stimuli is reflected in discrepancy theories of reward or reinforcement (Donahoe & Palmer, 1993, Berridge, 2004) that demonstrate how positive unexpected variances in the functional relationship between behavior and reward co-vary with changes in the activity of dopamine neurons, and code the difference between the expected and actual value of outcomes (Schultz, 1998). Unpredicted negative changes or counterfactual changes in the discriminative function of a stimulus (Camille et al. 2004, Zeelenberg, 1998) perform a similar function by also coding a difference between actual and expected values.

Discrepancy may also be perceived from an estimate of the relative or contrasting goodness of a response. Specifically, discrepant positive or negative variance may be perceived while performing under a primary response contingency, but also concurrently with a discrepant or surprising variance revealed by alternative feasible schedules of reward, as illustrated by the concept of behavioral contrast (Flaherty, 1996). This occurs when the goodness or badness of a reward is determined through its contrast to other response options that are revealed prior or subsequently to choice, and represent feasible opportunity losses or gains. The relative goodness of a response option may be a discrete event consciously perceived, or it may represent an indiscrete event that may be consciously or non-consciously perceived, and represents a ‘distractive’ loss. Distraction will be defined as a small or a series of small imminent feasible (i.e. avoidable) losses, consciously or non-consciously perceived (e.g., moment to moment feasible loss of access to internet or email while at work). In the many situations when a plenitude of options make their goodness difficult or impossible to logically calculate, choosing one option will more likely result in a subsequent interpretation of discrepant losses or gains when comparatively better or worse choices are revealed in hindsight. For example, choosing to wait in one check out line among several at a grocery store will result in tension when other lines move faster relative to yours, but will be rewarding and tension free if your line is the fastest.

Negative Discrepancy and Tension

As defined by Damasio (1995), the somatic marker represents a somatic alarm bell for ‘danger’ ahead. As reflected in his interpretation of the IGT, this element of danger may be defined as a negative overall outcome, and is uniquely signaled by negative surprising counterfactual events (i.e. bad card choices). But what is the ultimate signal for tension, the negative card choice alone or the negative value of the overall deck imputed by that choice? Empirical evidence strongly suggests the former. This is due to the fact that ultimate danger in many cases may be recognized yet not signal tension, and that negative counterfactual events are correlated with tension regardless of whether the overall outcome is negative. For example, positive counterfactual events that also signal danger have no correlation with tension. This premise is supported by empirical observations of behavior that result in not intermittent but continuous surprising positive counterfactual outcomes. For example, schedules of reinforcement that match demand with skill and thus entail continuous positive surprise or discrepancy (Csikszentmihalyi, 1990) in otherwise dangerous situations (e.g., rock climbing, extreme sports) are highly correlated with profound relaxation and low autonomic arousal. As an exception, a high magnitude positive surprise may be associated briefly with elevated SCR that acts as a reflexive startle or orienting response, as when an individual is startled due to the magnitude of surprise (e.g. laughter at a comedy show, joy upon winning the lottery), but this rapidly declines with the habituation of the novel event and is followed by a general reduction of SCR (Lang et al. 1990). On the other hand, reinforcement schedules that entail continuous negative discrepancy (e.g. minding a child, waiting on tables) and do not signal overall danger are associated with elevated autonomic arousal due to tonic levels of tension that is reported as stress or anxiety (Wursted et al. 1991, 1996; Hagg, 1991; Lundberg, 1999). The latter also holds true in situations of behavioral contrast, and particularly when the likelihood of discrepant loss prior or subsequent to choice increases as the number of choices increase or when choices cannot be rationally compared (Marr, 2006) such as in the case of affective vs. rational choices (e.g. comparing the pleasurable value of eating a donut vs. the rational value of minding a diet). In the former case, tension cannot predict overall value, and occurs independently of the overall outcome or value of the primary contingency. For example, choosing one pair of shoes among a myriad options at a clothing store will more likely result in tension upon the subsequent appraisal of options foregone or that will be foregone, and not because of the value of the original purchasing decision. This ‘tyranny of choice’ (Schwartz, 2004) has been repeatedly demonstrated by social psychological observations of the correlation of tension, anxiety, and stress with multiple choices.

The Role of Affect

The observation that the anticipation of negative counterfactual events or discrepancies is uniquely correlated with phasic (SCR) and tonic (SCL) muscular tension and associated autonomic arousal can also be described by discrepancy and tension (including of course the neuro-physiological systems they modulate) are denoted as not just informative events, but also as an equivalent hedonic or affective events. That is, discrepancy and tension not only denote information about means-ends expectancies but also represent painful or pleasurable affective states that in themselves modulate behavior in ways that may not be coherent with rational behavior. To understand affect, it is important to understand its semantics. It is an unremarkable fact in science that identical processes may have different corresponding and equally valid metaphorical representations (Lakoff & Johnson,1999). Thus aspects of vision may be represented by the activity of rods and cones in the retina and attendant neural processes, or may be more simply and no less accurately denoted as the perception of colors such as red, blue, and green. Similarly, the metaphorical representation of the molecular processes of tension and associated arousal can also be represented through the molar metaphor of aversive or painful states. This reported ‘feeling’ is defined as affect. More remarkable however is the fact that discrepant events are affective events as well, and are also reported as pleasurable or painful. Unpredicted positive and negative changes (Shepperd & McNulty, 2002) in the discriminative function of a stimulus respectively result in self reports of elation or depression, and the experience of negative discrepant events (e.g. embarrassment, regret) as well as negative affect derived from the peripheral nervous system (e.g. having a tooth pulled) also correlates with the affective event of muscular tension and corresponding approach or avoidance behavior (Mellers & McGraw, 2001, Miller, 1992). In addition, the modeling or ‘visualization’ of future positive or negative discriminative events may elicit an approximation of a response before it’s eliciting stimuli actually occurs. The elicitation of the sensory and motor elements of a response by this ‘anticipation’ is called priming (Donahoe & Palmer, 1993). Thus a student may feel euphoric or depressed in anticipation of a respective likelihood of a good or bad failing mark in a class, and psychotherapeutic outcomes rely upon changing patterns of thoughts to those that elicit positively affective emotional primes (e.g. optimistic thinking). Finally, as experience demonstrates, pain or pleasure scales with the relative importance and duration of a counterfactual event or events. Thus, bad news ‘hurts’ more and good news feels better as these events scale in magnitude. Also, as the Cinderella effect demonstrates, a continuous succession of small negative feasible counterfactual events (i.e., distractions) can cause severe tension induced pain and emotional distress as a distractive day at work or minding the children calls to mind.

Negative affect may be generated by the peripheral or central nervous systems, and originates from physiological (pain from muscular tension or physical trauma) or cognitive (negative discrepancy of ‘disappointment’) or distractive causes. However, the function of pain, regardless of its source, is generally regarded as not representational but imperative. That is, the function of pain is not to inherently represent the advisability of specific action plans but rather to interrupt and redirect attention, and impose a new action priority to escape (Eccleston & Crombez, 1999). That is, pain signals the molecular moment to moment goodness of an individual response rather than the molar goodness of a response set. Given this fact, it follows that the pain of tension logically serves the same function, which is not to help parse long term decisions (as per Damasio’s interpretation) but to set a new priority for short term decisions that lead to escape.

Although negative affect or the anticipation of negative affect correlates with the affective state of tension, the experience or anticipation of positive affect does not. Like pain, positive affect connotes molecular or moment to moment goodness rather than the long term or molar value of behavior. And as with pain, the momentary goodness may or may not cohere with the overall goodness of behavior. As mentioned earlier, in ‘flow experiences’ (Csikszentmihalyi, 1990) such as creative behavior that represent a succession of moment to moment positive discrepancies or surprises and attendant pleasurable affect, relaxation or low autonomic arousal is commonly reported. This positive affect gives value to momentary behavior independent of its long term or extrinsic results. Specifically, in humanistic and social psychology, positive affect elicited by moment to moment behavior represents the intrinsically reinforcing or ‘autotelic’ aspect of that behavior, where you do it for the sake of doing it, and has no bearing on the valuation of the rational or ‘extrinsic’ value of behavior and has no correlation with increased tension or anxiety (Deci & Ryan, 1995).

Finally, the motivational systems that mediate discrepancy based affect are different psychologically and neurologically from the cognitive act-outcome expectancies that mediates rational goal making (Berridge, 2001), and therefore it follows that affect does not inherently denote effect, but rather constitutes an unconditioned response to abstract properties of the contingency, or in other words is a schedule effect. That is, although positive affect may cohere with and enhance (or somatically mark) the importance of long term goals, this coherence is not implicit in affect but rather is an artifact of how discrepancy or novelty is rendered through the schedule of reinforcement. For example, piece work or fixed ratio schedules of reinforcement require reward to follow fixed and predictable performance before the onset of reward, and are generally regarded by the subject as boring or emotionally painful. However, if perceived progress under the same contingency is unpredictable, then performance is regarded as interesting and pleasurable, even if the long term results are known by the subject to be negative. For example, a teacher reinforces a child’s progress in reading through unpredicted praise, and the child’s positive feelings are coherent with successfully learning to read. But this reinforcement is due to the teacher making sure that praise and it’s accompanying positive affect is coherent with the long term goal of learning to read. Positive affect thus motivates good behavior because it is designed to do so, and not because it intrinsically denotes the overall correctness of choice. Likewise, a casino operator reinforces a gambler’s activity through a succession of small successive wins (which mitigate the occasional large loss), but also ensures that the gambler’s positive feelings are dis-coherent with winning over the long term. Similarly, an individual may adjust the difficulty of a game to provide enough positive surprise to make the game pleasing and thus motivating, or he may engineer a similar contingency by proverbially waiting until the last minute to do a task, thus insuring that positive surprise (and affect) will motivate him to get to work on time, complete his taxes, etc. Finally, in a variant of the IGT surnamed the Soochow Gambling Task or SGT,(Lin et al. 2007, Chiu et al. 2008), it was demonstrated that the choice of card deck was determined by the frequency of small positive variations in the reinforcement schedule during the performance, resulting in choice that was independent of the expected overall value of the performance outcome.

Thus it may be hypothesized that regardless of whether it is mediated by the central or peripheral nervous system or is positive or negative in nature, no affective event can independently determine or predict rationally conceived long term or molar value. That is, and contrary to Damasio’s hypothesis, so called gut feelings alone cannot predetermine the adequacy of long term decisions.

Originality and Limitations of the IGT

The IGT was designed to track overt and covert responses under alternative contingencies wherein reinforcement is variable or uncertain. Yet, measuring such choice-choice behavior under uncertainty is not new either in terms of the experimental paradigm or in terms of its dependent measures. Indeed, using the SCR to measure covert responding under similar conditions of choice provided the experimental foundation of the Dollard and Miller theory of anxiety (Miller, 1992). What is new is that the SCR was imputed to modulate the effectiveness of long term choice rather than short term avoidance, and that Damasio renders a sophisticated neurological rather than learning based explanation of the SCR. Another more serious problem is that the dependent measures of the IGT, namely the ‘gut level’ responses, are incomplete both for Damasio’s original experiment and its subsequent replications.
As derived from the IGT, the somatic marker hypothesis imputes that an affective ‘gut level’ somatic response (i.e., tension mediated autonomic arousal) results from the appraisal of surprising counterfactual outcomes or discrepancies that are implicit in choice. However, the immediate perception as well as anticipation (priming) of these outcomes is also affective, and this affect is mediated by the neuro-modulator activity (e.g., activity of midbrain dopamine systems) that modulates choice. Thus not one but two ‘gut level’ affective responses may occur in tandem across performance under the IGT. Specifically, neurally based affect as elicited by the perception of decision-outcome discrepancy or priming effects due to the anticipation of discrepancy is not addressed in the experimental measures of the somatic states that are induced by the IGT and its various experimental iterations. But this is not a limitation of the IGT but of the observational tools that may be applied concurrently with the experiment. As originally implemented, the IGT represents an experimental ‘snapshot’ of the covert and overt behavior concomitant with decision making or choice. However, in studies that replicate the IGT, only a subset use psycho-physiological data (Dunn, 2006 ), and these data generally use indirect measures of muscular tension as indicated by the SCR. The unique psycho-physiological affective responses due to the actual and anticipated perception of positive or negative discrepancy as a subject chooses cards are not measured or represented as affective states. Moreover, because tonic levels of muscular tension as reflected by the SCL are not considered, affective states that result from the tonic activation of the musculature such as stress are not revealed. These represent serious omissions in the experimental investigation of the somatic marker, since all the relevant affective events occurring due to choice are not considered. Specifically, in decision making peripheral and neurologically based affective responses often occur in tandem, yet no experiments have measured their correlation with the experimental contingencies that underlay choice. Thus the covert responses that mediate choice between alternative contingencies are measured separately through measures of the source of peripheral (e.g., muscular tension) based affect (e.g. EKG, SCR), or in-vivo brain imaging (Berridge,2004) that measure the source of neurologically based affect (e.g., activity of dopamine neurons). However, none have measured them together as events that co-vary across time and performance (i.e. through a within group experimental design characteristic of Skinnerian behavior analysis). The question is how covert and overt behavior under the IGT or in a more general sense alternative response contingencies or ‘decision making’ can be explained through an integration of the data provided by both of these experimental perspectives as reflected in the comprehensive discipline of affective neuroscience.

Assembling an Explanation

The lessons of the IGT experiment as informed by data regarding the affective neuroscience of choice are simple. First, the negatively affective (i.e. painful) states of tension and autonomic arousal correlate with and precede or attend the occurrence of a negative discrepancy that also is represented affectively as painful. Secondly, if pain has an imperative and not representational quality, then the somatic marker of tension mediates not effective decision making, but decisions that can facilitate avoidance or escape from the painful entailments of tension based arousal. Third, if tension is behavioral, then it and its opposite state of relaxation is dependent upon the arrangement of information as denoted by response contingencies.

These hypotheses have four major and testable entailments.

1. If anticipated pain elicits tension, then tension will occur regardless of the source of that pain.

Pain may originate with abstract information that is processed cognitively by cortically instantiated processes (perception, or thinking) or by input from the peripheral nervous system. For example, an individual may become tense in anticipation of finding out a likely poor course grade, or in anticipation of having a tooth pulled. Both are painful, and as common experience demonstrates, elicit tension based arousal when they are anticipated.

2 . If avoidance or escape cannot occur or will certainly occur, tension will not occur.

If avoidance or escape is perceived as impossible or impractical, loss will still occur but will be rendered non-feasible and will not correlate with tension. Non-feasible losses are defined as opportunity losses or beneficial response options or choices that by habit or circumstance cannot respectively be avoided or performed. For example, while at work we possess the option to leave to see a movie, fly to Paris, or go fishing, but habit or circumstance leads us to be unconcerned about their loss, or feasibility. However, when we are planning a vacation when a choice between such outcomes is incumbent, then choosing between similar response options does represent a cause for feasible loss and tension. Similarly, as perceived feasible loss becomes non feasible due to experience or ‘extinction’, tension will correspondingly decrease. Thus a student who is experiencing test anxiety becomes less anxious if avoidance is eliminated through a ‘time out’ (Gresham & Kern, 2004) that extinguishes tension by reducing the prospect of avoiding prospective loss. Although the student may still experience loss and the regret it entails, muscular tension declines when the student realizes it cannot be reinforced. Similarly, individuals who perceive likelihood of a significant loss of life or property (e.g. a hurricane or disease) will feel tense when there is still a prospect that the loss will be avoided, and merely regretful or depressed when they know it cannot. Also, when avoidance from physical pain is impossible, as in the case of ‘learned helplessness’ (Seligman, 1975; Gatchel et al. 1977), low autonomic arousal is characteristic.
Finally, when avoidance is certain, tension will also not occur. For example, whereas an inescapable electric shock renders the subject (in this case an experimental animal such as a dog) (Seligman, 1975) ‘helpless’ and relatively anxiety free, completely avoidable painful events (e.g. avoiding a dangling power line seen from a distance) are similarly tension free. Similarly, game animals are nonchalant when they knowingly keep a safe distance from predators that are seen and are thus predictable in the distance.

3. The absence of feasible loss will be accompanied by a lack of tension, or relaxation, independent of the level of task demand.

In commonly accepted explanations for tension or stress (Selye, 1980), the source of stress is any type of demand represented by not physical but cognitive operations. That is, much as physical demands (e.g. running, exercising) cause ‘wear and tear’ on the body, cognitive demands (continuous cognitive appraisal or thinking) cause a similar wear and tear represent by systemic autonomic arousal or ‘stress’. Ostensibly, this has been confirmed by well known and amply tested relaxation protocols such as meditation and mindfulness meditation (Kabat-Zinn, 1993) that presume that relaxation is induced by a complete avoidance of complex cognition. Yet in this literature, only recently have studies begun to investigate feasible (i.e., distractive loss) loss as a separate qualitative aspect of cognition that may be used as an independent measure of tension and relaxation. These preliminary findings have lead to the observation that the ability to reduce distraction is a ‘unique mechanism by which mindfulness meditation reduces distress’( Shamini et al., 2007). But overall, in meditative disciplines the separate emotional import of feasible and non-feasible loss are not differentiated, leading to the unproven and unjustified notion that relaxation can occur specifically through the reduction of complex cognition or high cognitive ‘demand’ rather than an aspect of cognition or demand.

In fact, complex cognition often occurs with relaxation, and relaxation is necessary to optimize cognition. Indeed, effective thinking more often has a negative rather than positive correlation with tension. For example, declarative reasoning is enhanced during states of rest (Grecius et al. , 2003, Raichle et al., 2001), thus leading to the opposite conclusion that tension is deleterious to effective thinking. Moreover, proprioceptive feedback from the musculature is not required for problem solving (Rolls, 1999, Taub et al. 1966, Teuber, 1972), and executive attention and conscious feeling are not interdependent but are dissociable events (Naccache et al. 2004). In a critical overview of the relevant literature on Damasio’s hypothesis, Dunn concluded (2006) that peripheral feedback and associated arousal does not reliably influence decision making accuracy, and may reflect the end product of decision making rather than a key feature in its development. Finally, although it is commonly expected that tension will occur in high demand cognitive activities, this is not the case when feasible loss is absent. During high demand activities (e.g. creating art, climbing mountains) that perfectly match with skill and hence entail no feasible loss, low arousal or muscular relaxation is commonly reported (Perry, 1997).

4. Arousal is not unitary.

Arousal is a theoretically incoherent term, as it may apply to distinctive neuro-physiological processes (Robbins, 1997) that have equally distinctive causes. Yet, in contemporary accounts of performance motivation and arousal (Easterwood, 1959, Damasio, 1995), and indeed all contemporary commentary and experiment on the somatic marker hypothesis, these processes are not separated or controlled for, resulting in the confounding of their differing respective causes and effects. Thus, although neurologic arousal as marked by heightened alertness may or may not occur concurrently with the somatic response of tension and autonomic arousal, both are commonly combined and used as a practically indivisible state of arousal. For example, attentive arousal felt when watching a football game or riding a roller coaster often occurs concurrently with tension based arousal, forming a species of stress or ‘eustress’ from the conflation of two psychologically and physiologically distinct events. Similarly, an individual may compare multiple positive surprising outcomes, and may be tense (or excited) due to the contrasting choices, yet still be positively aroused due to the positive outcome itself. Thus, his subjective excitement or ‘joy’ describes two separate arousal states that are induced by two separate causes. Indeed, the classic inverted U curve (Yerkes & Dodson, 1908, Easterbrook, 1959) correlating the relationship between performance and arousal is better explained through the recognition that arousal in incentive motivation may represent not one but two separate dependent variables that co-vary with different independent causes.

For example, if task performance is plotted across the level of demand, the likelihood of moment to moment discrepant gain would at first offset the likelihood of loss, a circumstance that would reverse itself as demand increases. Thus, as a function of increasing demand, alertness and performance increases, but eventually decreases with a corresponding rise in tension due to an increased rate of ‘bad news’ implicated by high demand. Yet high demand does not necessarily implicate difficult or unattainable performance. Indeed, as earlier argued, remove the negative but not the positive discrepancy entailed by moment to moment performance and retain the demand for performance, and tension and autonomic arousal will not increase with demand, and high demand will correlate with a steady state of relaxation and low autonomic arousal. In addition, performance will not trail off and deteriorate with demand, but increase, and corresponding affect will be positive and quite pleasurable. This resulting ‘flow’ experience (Csikszentmihalyi, 1990) thus demonstrates how escalating demand increases alertness and performance, yet when discrepancy is controlled for, increasing demand will not result in increased autonomic arousal (or stress) and decrements in performance as the Yerkes-Dodson model predicts.

A New Synthesis

In sum, tension induced autonomic arousal occurs to somatically mark future uncertainly avoidable negative affective events that in turn originate from the peripheral (pain, as in dental office) or central nervous systems (cognitive discrepancy, such as losing or anticipating losing one’s job, or consistent distractions at work). Because it is affective, tension occurs and is reinforced by any decision that results in avoidance of future pain or escape from present pain, and not decisions made by other normative or rational criteria. That is, as painful events, tension and arousal are not representational, but imperative. Tension marks the anticipated goodness of moment to moment or molecular responding rather than long term or molar ends.

Unlike priming responses (e.g. positive affect due to thinking optimistic thoughts) which are elicited independently of action, the affective response of tension is emitted because it leads to action (namely successful avoidance), and thus is a conditioned response, or is behavioral. Relaxation is thus predicted to be coextensive with the consistent avoidance of feasible loss. Finally, although tension occurs concomitant with choice, periodic choices (e.g. choosing an item from a dinner menu), allow the nervous system to rapidly return to a homeostatic resting state. However, if choice occurs from moment to moment (e.g. continuous choosing during the working day between surfing the internet or working), then as the Cinderella effect predicts, tension will be tonically sustained and result in the physiological and emotional concomitants of stress. In other words, although phasic neuro-muscular activity occurs with periodic choice, tonic neuro-muscular activity is a function of continuous or moment to moment choice; and it is the latter that is commonly associated with the classic symptoms of stress.

This position that tension mediates avoidance conforms with Zajonc’s (1998) argument that emotions are designed to help individuals make approach-avoidance distinctions, and more specifically, that tension based arousal or ‘anxiety’ (Mowrer, 1939, Miller, 1992) is reinforced through its mediation of the avoidance of future painful events. This position contrasts with the true-false determinations that are mediated by Damasio’s somatic marker (Damasio, 1995). Finally, the fact that tension embodies an affective quality distinguishes the hypothesis from the computational role assigned by Damasio. That is, tension does not mark value because it is perceived, but rather it marks value because it is perceived to hurt. This affective value of tension and tension induced arousal serves as a mechanism to interrupt and redirect cognitive processing by bypassing rather than enhancing cognitive filtering and thus expediting action leading to avoidance or escape (Lowenstein et al., 2001, Armony et al., 1997). It is through this facilitation of action that tension is reinforced.

The presumption that tension is instigated by choice and is reinforced by successful avoidance also conforms with social psychological findings regarding the rationality of decision making. As commonly conceived, tension based arousal is mediated by stimulus-response (S-R) (Seyle, 1980) or response-outcome (R-S) (Damasio, 1995) processes, and tension respectively diminishes with successful avoidance (i.e. flight or fight) or when decisions are successfully parsed or made. In the former perspective, tension is an instinctive reaction to demand, and in the latter perspective, tension is a learned response that is reinforced because it facilitates long term decision making. But social psychological data do not support either of these positions. In fact, the experience of tension based arousal is mediated by choice, not demand. Secondly, tension induced by multiple decisions or choices mediates not effective choice but the avoidance of choice, and does not serve formal principles of rationality (Keys & Schwartz, 2007).

As a final note, it must be remarked that this conceptualization is not new, and returns the analysis of tension to the seminal concepts of Sigmund Freud. Specifically, Freud’s concept of ‘signal anxiety’ (Wong, 1999) posited like Damasio that tension based arousal unconsciously signaled ‘danger’ ahead. But in contrast to Damasio, the results do not make for better choices, but are a cause for avoidance of overt or covert (i.e., repressing thoughts about an object) behavior. That is, tension results in the avoidance of the momentary pain of tension rather than informing the goodness of long term behavior. Subsequently, in a series of experiments involving choice utilizing the SCR as the dependent measure for tension, Neal Miller (1992) reframed the Freudian role of anxiety within a learning perspective, and the resulting Dollard and Miller hypothesis for anxiety posited that the affective (i.e. painful) event of tension occurs in anticipation of physically or emotionally painful events and is reinforced by successful avoidance. Ironically, the Dollard and Miller hypothesis did not gain prominence in large measure because of its simplicity (Bolles, 1994), a salient characteristic that it shares with the equally simple but much more notable hypothesis of the somatic marker.

Tension and Stress

As derived from the IGT experiment, the concept of tension as a somatic marker forces us to rethink the nature of the covert state of muscular tension as not a reflexive but a behavioral or conditioned response. Yet, in contradistinction to Damasio’s hypothesis, tension and autonomic arousal change the value not of long term but of moment to moment responding, which may or may not cohere with rational long term goals. In addition, by attributing tension and accompanying arousal to the perception of feasible and affective losses both large and small (i.e. distractive losses), tension becomes a function not of demand but of the feasibility of the avoidance of discrete stimulus events. Thus to be relaxed requires a complete revision of what relaxation entails and its opposite state of tension based arousal or stress.

In his class, the psychologist F. J. McGuigan (1993) would induce relaxation in his students through the technique of progressive relaxation. He would then drop a book to demonstrate how the startle reflex and related tension and associated arousal is inhibited or impossible without the presence of muscular tonus, a finding originally made by Sherrington (1909) and explained neurologically by Gellhorn (1967,1972). However, although all remedies for stress posit relaxation as the necessary antidote for and indeed opposite of stress, the dependent measures for stress in almost all academic opinion represents various measures of arousal as reflected in self reports and physiological and neurological indices (e.g. GSR, EEG). But by confusing the myriad subjective and objective symptoms of stress with its uniform neuro-muscular source, the concept of stress is rendered theoretically incoherent, since there is no one consistent dependent or independent measure for stress that all parties can agree. The result is that stress becomes merely a taxonomy for all non-specific arousal caused by non-specific stimuli categorized by ‘demand’ rather than a clearly defined behavioral response, and renders a panoply of cures that are in some respects worse than the disease.

Before the advent of modern medicine, disease was thought to be caused by an overabundance of impurities or ‘plethoras’ in the blood, and was treated by nostrums such as blood letting that generally caused more harm than good. It was only when illness was redefined by metaphorical representations of discrete viral and bacterial causes that allowed for a scientifically valid therapeutic focus on the prevention and treatment of disease. Similarly, the governing metaphor of tension or stress is that it is a product of an abundance or plethora of demand (Selye , 1980) and that relaxation is a function of reducing the daily demands of life, by means of vacations, pursuing non demanding diversions at home or work, or through procedures such as meditation or relaxation training that withdraw an individual from the demands of the world. Yet by adding continuous feasible choices in the place of demanding ones, we increase also feasible losses. Thus we retreat from a world of complexity to a world of ever proliferating distractive choices, and end up as the Cinderella effect suggests in a state of perpetual tension and emotional exhaustion. Furthermore, by eliminating all demand rather than controlling for aspects of demand, relaxation becomes a mere product of a cloistered environment, and not an accompanying trait of an active and demanding lifestyle. Specifically, relaxation is not a function of radically reducing thought (e.g. mindfulness meditation), but of radically reducing distraction. That is, radically reducing distractive judgements or choices is not incidental to the induction of relaxation, it is fundamental.

Finally, the methodology of learning theory (as exemplified by experimental designs such as the IGT) allows for the analysis of the fine grain co-variations between stimuli and behavior for single subjects mapped across time (i.e. a ‘within’ group experimental design), and can uniquely demonstrate how practice variables or experience modulate tension and predict its future occurrence. Such co-variations have generally not been considered as independent measures of tension either experimentally or descriptively, and are subsumed under molar concepts such as ‘attention’ or ‘demand’ that can only be imperfectly described through the use of literal language. This is reflected in the many disparate and conflicting definitions of stress. Learning theory mitigates this problem by establishing a much greater precision in the description of the independent variables that correlate with tension and stress. Specifically, if the fine grain or molecular metaphor of ‘choice’ replaces the molar metaphor of ‘demand’ as the primary descriptor of the etiology of tension, then the management of simple contingencies of reinforcement are implicated in a much more precise and uniform description of the origin, prediction, and control of the operational measures of everyday stress. But this avenue has scarcely been researched due in no small measure to the rejection or neglect of the experimental methodology of learning theory as a necessary tool for the analysis of stress. Nonetheless, this argument is won not by the parsimony and precision of a learning based explanation, but through the power of procedure to effect behavioral change. That of course is the mandate and justification of a true science of behavior.


Armony, J. L., Servan-Schreiber, D. Cohen, J. D. & LeDoux, J. E. (1997) Computational modeling of emotion: Explorations through the anatomy and physiology of fear conditioning. Trends in Cognitive Sciences. I. 28-34

Bechara, A. & Damasio, A. (2005) The Somatic marker hypothesis, a neural theory of economic decision. Games & Economic Behavior, 52, 336-372

Bolles, R. (1994) The Story of Psychology: A Thematic History. Belmont, Ca.: Brooks/Cole

Berridge, K. (2004) Motivational concepts in behavioral neuroscience. Physiology and Behavior, 81, 179-209

Berridge, K. (2001). Reward learning: reinforcement, incentives, and rxpectations, The Psychology of Learning and Motivation, 3, Academic Press, New York.

Camille, N. Coricelli, G. Sallet, J. Pradat-Diehl, P., Duhamel, J. & Sirigu, A. (2004) The involvement of the orbitofrontal cortex in the experience of regret. Science, 304(5674), 1167-1170

Chiu, Y., Lin, C., Huang, J., Lin, S, Lee, P., & Hsieh, J. (2008) Immediate gain is long term loss. Are there foresighted decision makers in the Iowa Gambling Task? Behavioral and Brain Functions, 4(13) (online)

Csikszentmihalyi, M. (1990) Flow, the psychology of optimal experience. New York: Harper Collins.

Damasio, A. (1995) Descartes Error: Emotion, Reason, and the Human Brain. Avon: New York

Deci, E. C., & Ryan, R. M. (1995) Intrinsic motivation and self determination in behavior, New York: Springer

Donahoe, J.W., & Palmer, D. C. (1993) Learning and Complex Behavior. Needham Heights, Ma: Allyn and Bacon

Dunn, B. D., Dalgleish, T., & Lawrence, A.D. (2006) The somatic marker hypothesis: A critical evaluation. Neuroscience and Biobehavioral Reviews, 30(2): 239-271

Easterbrook, J. A. (1959) The Effect of Emotion on Cue Utilization and the Organization of Behavior. Psychological Review, 66, 183-201

Eccleston, C. & Crombez, G. (1999) Pain demands attention: a cognitive-affective model of the interruptive function of pain. Psychological Bulletin, 125(3): 356-366

Flaherty, C. (1996) Incentive Relativity. New York: Cambridge

Gatchel, R. J., McKinney, M.E., Koebernick, L. F. (1977) Learned helplessness, depression, and physiological responding. Psychophysiology, 14 (1), 25–31.

Gellhorn, E. (1967) Principles of autonomic-somatic integration. Minneapolis: University of Minnesota Press

Gellhorn, E. & Kiely, W. F. (1972) Mystical states of consciousness: Neurophysiological and clinical aspects. Journal of Nervous and Mental Disease, 154, 399-405

Greicius, M. D. Krasnow, B., Reiss, A. L., Menon, V. (2003)Functional connectivity in the resting brain: A network analysis of the default mode hypothesis. Proceedings of the National Academy of Science, 100, 253-258

Gresham, F. G., Kern, L. (2004) Internalizing Behavior Problems in Children and Adults. In Rutherford, R. B., Quinn, M. M., Mathur, S. R. (Eds.), Handbook of Research in Emotional and Behavioral Disorders. New York: Guilford

Hagg, G. (1991). Static Work loads and occupational myalgia- a new explanation model. In P. A. Anderson, D. J. Hobart, and J. V. Danhoff (Eds.). Electromyographical Kinesiology (pp. 141-144). Elsevier Science Publishers, P. V.

Jacobson, E. (1970) Modern treatment of tense patients. Springfield, Il: Charles C. Thomas.

Kabat-Zinn, J. (1993) Mindfulness Meditation: Health Benefits of an Ancient Buddhist Practice. In Goleman, D. and Gurin, J. (eds.) Mind/Body Medicine, Consumer Reports Books, Yonkers, Ny.

Keys, D. J. & Schwartz, B. (2007) ‘Leaky Rationality’ How research on behavioral decision making challenges normative standards of rationality. Perspectives on Psychological Science. 2(2), 162-180

Lakoff, G. & Johnson, M. (1999) Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought, Basic Books

Lang, P. J., Bradley, M. M., Cuthbert, B. N. (1990) Emotion, attention, and the startle reflex. Psychological Review, 97(3), 377-395

Lin, C., Chiu, Y. Lee, P. & Hsieh, J. (2007) Is deck B a disadvantageous deck in the Iowa Gambling Task? Behavioral and Brain Functions, 3(16)(online)

Lowenstein, G. F., Weber, E. U., Hsee, Christopher, K. , & Welch, N.(2001) Risk as feelings. Psychological Bulletin, 127(2), 267-286

Lundberg, U. (1999) Stress Responses in Low-Status Jobs and Their Relationship to Health Risks: Musculoskeletal Disorders. Annals of the New York Academy of Sciences, 896, 162-172.

Malmo, R. B. (1975) On emotions, needs, and our archaic brain. New York: Holt, Reinhart, and Winston.

Marr, A. J. (2006) Relaxation and Muscular Tension: A Biobehavioristic Explanation, International Journal of Stress Management, 13(2), 131-153

McEwen, B. S. (1998) Stress, adaptation, and disease: allostasis and allostatic load. New England Journal of Medicine, 338, 171-179

McGuigan, F. J. (1991) Calm down: A guide for stress and tension control. (rev. ed.) Dubuque, IA: Kendall/Hunt

McGuigan, F. J. & Lehrer, P. (1993) Progressive Relaxation, Origins, Principles, and Clinical Applications. In Paul M. Lehrer (Ed.). Principles and Practice of Stress Management, 2nd ed. Guilford Press

Mellers, B. A., McGraw, A. P. (2001) Anticipated emotions as guides to choice. Current Directions in Psychological Science. 10, 201-214

Miller, N. (1992) Studies of fear as an acquirable drive: I. Fear as motivation and fear-reduction as reinforcement in the learning of new responses. Journal of Experimental Psychology: General, 121(1):6-11.

Mowrer, O. H. (1939) A stimulus-response analysis of anxiety and its role as a reinforcing agent. Psychological Review, 46:553-565.

Perry, J. T. (1997) In the Zone. Chicago, Ill.: Contemporary Books

Raichle, M. E., MacLeod, A. M., Snyder, A. Z., Powers W. J., Gusnard, D. A., Shulman, G. L. (2001) A default mode of brain function. Proceedings of the National Academy of. Sciences. 98, 676-682

Robbins, T. W. (1997) Arousal systems and attentional processes. Biological Psychology, 45, 57-71

Rolls, E. (1999) The brain and emotion. Oxford: Oxford University Press

Schultz, W. (1998) Predictive reward signal of dopamine neurons. Journal of Neurophysiology, 80, 1, 1-27

Schwartz, B. (2004) The Paradox of Choice: Why More is Less. Hopewell, N. J.: Ecco.

Seligman, M. E. P. (1975) Helplessness: On depression, development and death. Freeman, San Francisco.

Selye, H. (1980) Selye’s Guide to Stress Research, New York: Van Nostrand Reinhold

Sherrington, C. S. (1909) On plastic tonus and proprioceptive reflexes. Quarterly Journal of Experimental Psychology, 2, 109-156

Shamini J., Shapiro, S. L., Swanick, S., Roesch, S. C., Mills, P. J., Bell, I., Schwartz, G.E.R., (2007) A Randomized Controlled Trial of Mindfulness Meditation Versus Relaxation Training: Effects on Distress, Positive States of Mind, Rumination, and Distraction. Annals of Behavioral Medicine, 33(1), 11-21

Sheppard, J. A. & McNulty, J. K. (2002) The affective consequences of expected and unexpected outcomes. Psychological Science, 13, 85-88

Taub, E., Ellman, S. J., & Berman, A. J. (1966) Deafferentation in monkeys: effect on conditioned grasp response. Science, 151, 593-594.

Teuber, H. L. (1972) Unity and diversity of frontal lobe functions. Acta Neurobiologiae Experimentalis, 32, 615-656.

Tomb, I, Hauser, M., Deldin, P. & Caramazza, A. (2002) Do somatic markers mediate decisions on the gambling task? Nature Neuroscience, 5, 1103-1104

Wong, P.S. (1999). Anxiety, signal anxiety, and unconscious anticipation: Neuroscientific evidence for an unconscious signal function in humans. Journal of the American Psychoanalytic Association, 47, 817-841

Wursted, M., Bjorklund, R., & Westgaard, R. (1991) Shoulder muscle tension induced by two VDU-based tasks of different complexity. Ergonomics, 23, 1033-1046

Wursted, M., Eken, T., & Westgaard, R. (1996) Activity of single motor units in attention demanding tasks: firing pattern in the human trapezius muscle. European Journal of Applied Physiology, 72, 323-329

Yerkes, R. M. , & Dodson, J. D. (1908) The relation of strength of stimulus to rapidity of habit formation. Journal of Comparative Neurology and Psychology, 18, 459-482

Yucha, C.B., & Gilbert, C. (2004) Evidence Based Practice in Biofeedback and Neurofeedback. Monograph Wheat Ridge, CO: Association for Applied Psychophysiology and Biofeedback

Zajonc, R. (1998) Emotions. In D. Gilbers, S. Fiske, & G. Lindzey (Eds), Handbook of social psychology, New York: Oxford University Press

Zeelenberg, M. , van Dijk, W., van der Plight, J. Manstead, A., van Empelen, P. Reinderman, D. (1998) Emotional reactions to the outcomes of decisions: The role of counterfactual thought in the experience of regret and disappointment. Organizational Behavior and Human Decision Processes, 75, 117-141

No comments: