Search This Blog

Thursday, January 04, 2007

The Happy-Unhappy Little Curve of Yerkes and Dodson


In psychology, nothing is more impressive than a bell curve. Bell curves tell you where you stand academically, socially, and psychologically. They are a swell way of graphically making an argument that hedges its bets. Thus you can be anywhere on the curve, it just depends. Bell curves also have a faux mathematical rigor about them. Like a physical law, you change one variable and the other one changes in a proportional way. One such behavioral algorithm is the Yerkes-Dodson curve, which is more metaphor really, as we commonly invoke the Yerkes-Dodson curve to support the hoary cliche in psychology that demand (i.e., stress) is good for you up to a point when things start going rapidly downhill (hence the bell curve).

The original Yerkes and Dodson published their hypothesis way back in the year 1908, and actually has little to do with the little graph you see above, which in turn doesn't have a lot of empirical support in psychology anyways, but I digress. Basically, the Yerkes Dodson curve plots performance against physical arousal, which presumably represents real discrete events that can be plotted across the X and Y axes. Thus given an X amount of performance, you can reliably infer a Y amount of arousal, and vice versa. This is all well and good if performance and arousal are consistently defined things. The problem is, for arousal at least, it's not. What is arousal? Indeed, there are many kinds: sexual, emotional, physical. Thus a fellow can be aroused while peeping into the girls locker room, and aroused in a different way upon being discovered, and aroused more differently yet as he hightails it away.

Moreover, different states of arousal have different relationships to performance, and can occur separately or at the same time. Attentive alertness, as a form of arousal, increases performance as arousal increases. On the other hand tension and attendant autonomic arousal, or anxiety, always decreases performance (see tomorrow's post for more). Interestingly enough, separate them both and the Yerkes-Dodson curve disappears, but combine them and out it pops. For example, a person who is highly and pleasurably aroused while climbing a mountain or creating art doesn't suffer in performance as his arousal increases, but actually gains in performance. On the other hand, a person who is frustrated while performing a task progressively loses his ability to perform well as anxiety increases. Nonetheless, as demand increases and decreases, these two very different types of arousal can occur simultaneously, and produce a performance curve very similar to the Yerkes-Dodson model.

As an aroused state, attentive alertness scales with the novelty or surprise of moment to moment behavior. As a function of the release of the neurochemical dopamine, touch and go events that entail continuous positive surprises (e.g. rock climbing, gambling, creative behavior) positively correlate with aroused alertness, which not only feels good but helps you think better. Thus the bigger the positive surprise, the more alert you become, and the better your performance becomes. If however, surprises start to trend from good to bad, alertness decreases as we become progressively more depressed, but tension and associated autonomic arousal (i.e. anxiety) increases. That is, as news moves from good to bad, arousal doesn't increase, it just changes to an entirely new form! The problem though is that positive surprises always come at the risk that things will take a decided turn for the worse, as the rock climber get stuck in a snow storm and the creative artist hits a writers block. Thus, the cost of higher good feelings is the chance you take that a turn of fortune will turn those good feelings bad. Generally, as demand increases risk increases, and at first we can handle it and be pleasantly surprised by and are motivated by the continuous moment to moment surprise of our success. But as demand ratchets up we are more likely to experience failure, and another type of arousal, that of anxiety. Hence as demand goes up, so does performance and arousal until performance reaches a crest and arousal begins to change not in amplitude but begins changing in kind. So the Yerkes Dodson bell curve survives, it is rather the idea that arousal does not change in kind across the level of performance that falls away.

The lesson we learn from all this is that the highest motivation or performance stands at the cusp of failure, as we are rarely motivated by the sure and thus boring thing. Unfortunately, what psychologists take from the Yerkes-Dodson curve is the wrong lesson entirely, that arousal is a monolithic and indivisible thing that does not categorically change as demand increases. In other words, the lesson that no pain equals no gain is wrong. Rather, if you have pain you will likely have no gain. For folks that are a bit wary of the school of hard knocks, this is perhaps a lesson one can get a bit excited about.


2 comments:

Anonymous said...

What do you think of more theoretically grounded takes on stress such as hancock's extended U model

A. J. Marr said...

Hancock is on target with his criticism of the a-priori acceptance of arousal as a unitary thing, but his theory still does not define arousal. So, philosophically we agree, but empirically his work still lacks a specific account of how arousal is engendered in situations that demand performance, or incentive motivation if you will. I am attaching an excerpt from my paper, the "The Operant of Tension" (available on google) that I hope can illustrate this point.


.... incentive motivation engages different neural and neuro-muscular processes that can be respectively defined as ‘arousing’, but represent separate and independent psychological and physiological processes. Indeed, ‘arousal’ is a theoretically incoherent term, as it may apply to distinctive neuro-physiological processes (Robbins, 1997) that have equally distinctive causes. Yet, in classical and contemporary accounts of performance motivation and arousal (Yerkes & Dodson, 1908, Easterwood, 1959, Damasio, 1994), 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. 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 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, 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 demand itself 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 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