Thursday, February 22, 2007

The Role of Somatic Markers in Trading Decisions

I think this is one of my more important articles: a perspective on trading psychology from the field of cognitive neuroscience.

In past posts, I have emphasized the importance of controlling the body's level of arousal as a fundamental trading skill, because so many psychological disruptions of trading are state-dependent. There are cognitive and behavioral methods for achieving such control; these include simple exercises that traders can perform on their own.

There is far more to trading psychology, however, than simply taming emotions. The idea that emotions are the root of trading problems--and that successful traders eliminate their emotions--is one of the great myths in the field. Indeed, research in cognitive neuroscience--particularly the work pioneered by Antonio Damasio--suggests that emotion is absolutely crucial in decision making.

Damasio's seminal contribution in this area of research is his somatic marker hypothesis. An overview of this hypothesis and relevant research is available in this review article. According to that article, the gist of the somatic marker hypothesis is that:

"Structures in ventromedial prefrontal cortex provide the substrate for learning an association between certain classes of complex situation, on the one hand, and the type of bioregulatory state (including emotional state) usually associated with that class of situation in past individual experience. The ventromedial sector holds linkages between the facts that compose a given situation, and the emotion previously paired with it in an individual's contingent experience. The linkages are ‘dispositional’ in the sense that they do not hold the representation of the facts or of the emotional state explicitly, but hold rather the potential to reactivate an emotion by acting on the appropriate cortical or subcortical structures..."

Let's paraphrase that in plain English! A section of the brain's frontal cortex is responsible for storing and processing associations between situations and the emotions that had been activated by those situations. When we store these associations, we are not storing conscious memories. Instead, we create links between situations and how those situations make us feel. Those linkages are "dispositional", which means that they lead us to act in one way or another.

Here's a simple example: I have many fond memories of spending time in cafes in the U.S. and Europe. If I pass a cafe on the streets of Seattle and smell the aroma of coffee, that perception activates the feeling-states associated with my prior experience. Thanks to those feeling states, I am disposed to enter the cafe and spend some time there. My decision to enter the establishment is not a simple, rational process of toting up pros and cons. Rather, the feeling state mixed with the perception of the cafe--a link embedded in the ventromedial prefrontal cortex--leads to the decision.

So it is with all decisions, Damasio argues. Feelings--from explicit emotions to felt body states--are joined with perceptions to guide our behavior.

The research cited in the review article suggests that, when people experience damage to their ventromedial prefrontal cortex, they are no longer able to make sound decisions. One especially fascinating lines of research involves a risk-taking game that requires decision-making similar to that involved in trading. This "gambling task" involves choosing cards from four decks. Two of the decks have higher payoffs on wins, but also larger penalties for losses, so that the overall returns are negative. Two of the decks have smaller payoffs for wins, but smaller penalties, and an overall positive expectation of gain.

At the start of the gambling task, subjects don't know the odds for the piles. Over time, however, normal subjects learn to prefer the piles with the positive odds for winning. When they are hooked up to biofeedback equipment, they display clear skin conductance responses ahead of making their choices. These stress-related feeling states appear to help guide the choices of piles. Conversely, subjects with damage to their ventromedial prefrontal cortices do not display such anticipatory responses. They lack the feelings that guide good decision-making. As a result, they display a sustained preference for the high-return, high-penalty, negative expectation decks. Without access to their encoded feeling states, they cannot make good decisions under conditions of risk and uncertainty.

Now here's the really interesting part. Normal adults who describe themselves as risk-takers do not have damage to their ventromedial prefrontal cortex regions, but they do display a similar tendency to persistently choose cards from the risky, low-return piles. When connected to biofeedback equipment, the risk-takers *do* display emotional signals prior to making their choices, not unlike the normal subjects. They tend, however, to override those signals with their explicit thought processes. The brain-damaged subjects, on the other hand, never receive the emotional signals from the ventromedial prefrontal cortex in the first place.

The implications for trading are significant. The idea that successful traders overcome or eliminate their emotions is completely off-base. If that were to truly happen, the trader would behave like a brain-damaged patient. Rather, it is the overriding of emotional signals and signals from felt bodily experience that facilitates poor decision-making. This overriding can occur because of anxiety, greed, or myriad cognitive rationalizations. The implication, however, is that--at some level--experienced traders receive valuable physical and emotional signals to guide their decision-making. It is the temporary lack of access to these signals that leads traders to behave like the risk-takers in Damasio's experiments.

In a recent post, I emphasized the value of biofeedback in achieving self control. It may well be, however, that the greatest value of such disciplines as meditation, relaxation training, and biofeedback is to help people maintain a clear mind so that they do not override and obscure the somatic markers that are necessary for sound decision making. I strongly suspect that traders would be better served by disciplines that enable them to be accurate observers of their feelings than prescriptions for squelching those feelings. Choice lies at the heart of trading, and if Damasio is correct, our felt experience is a necessary substrate of choice. The somatic markers are there; it's simply a matter of keeping our access open to them, even as we're surrounded by risk, reward, and uncertainty.


NO DooDahs said...

Seykota's Trading Tribe, anyone? This sounds very similar to his TTM.

Brett Steenbarger, Ph.D. said...

Hi NO DooDahs,

Yes, there is an interesting overlap here. When I last conversed with Ed Seykota (which was quite a while back via email), he seemed less than smitten with my emphasis upon biofeedback and cognitive neuroscience. Much of what he does comes out of the humanistic tradition of psychotherapy, including Carl Rogers, Fritz Perls, Alvin Mahrer, and Leslie Greenberg. The idea of that approach is to engage in experiential exercises to heighten one's awareness and resolve conflicts and blocks to growth.

My own emphasis is less on such "growth" work and more on skill development via biofeedback and other disciplines. But the two definitely converge on the notion that raw human experience--from explicit emotions to subtle body shifts--contain information that is valuable for decision making. Thanks for the observation.


F. said...

Yes, I have noticed quite often that when I am intensely concentrated on trading action with an open position, all of a sudden I will get a very bad feeling out of the blue about the current risk:reward and start reducing the position until I feel more comfortable with the exposure. I got the same feeling this morning as NQ made new highs and took off my entire position within 1 point of the top.

It's difficult sometimes to distinguish between the fear of losing paper profits and this emotional tell. It seems that position size tends to increase the likelihood of it being the former.

Being a former quant, it is difficult many times to pull the trigger when I cannot state my reasons for action explicitly.

AnaTrader said...

Hi Brett

Often we are told that to be a good trader, one must not be emotional.

The implication, however, is that--at some level--experienced traders receive valuable physical and emotional signals to guide their decision-making. It is the temporary lack of access to these signals that leads traders to behave like the risk-takers in Damasio's experiments. - Unquote.

Emotional in this context is, therefore, conducive to good trading.

Brett Steenbarger, Ph.D. said...


Clearly, not all emotional reactions are satisfactory guides to action. One of the problems is that, as a whole, people are not very attuned to their emotional lives and have difficulty sorting out their emotional experience. If you think of emotions as having tastes (like food), then the taste of felt knowledge is very different from the taste of worry/fear. Knowing those flavors is essential to sorting out what you can trust and what you can't--


D TradeIdeas said...

We referenced this great article in our blog today:

"Many pundits tell their audience that emotions in trading adversely affect performance and their P&L. Put a lid on them and remove them entirely they say. Others like Dr. Brett Steenbarger convincingly argue that they are essential to long term, consistent success of a trader. However you decide, we think the real weak link in the Apprentice's Boardroom that gets fired is a trader's opinions. Opinions get a trader in trouble - facts, probabilities, and statistics do not.

Here's my example: Think of the Oscars that are upon us this weekend. Will Scorcese win his FIRST Oscar for The Departed?! Should he have won for Casino? Goodfellas? Your answer, no matter how researched, rational or logical, is your opinion. What you think should happen is irrelevant. Keep those kind of opinions to the movies - not your trading. You'll be richer for it. The kind of tools and thinking to bring to your trading is what the LCS will be all about."

Our bottom line: Tested systems count, opinions don't

Brett Steenbarger, Ph.D. said...

Thanks for the comment, David. It's interesting to see how traders can become wedded to opinions, blinding them to a sensitivity to what's actually happening in the market: A malady I've encountered a few times in my career! Coordinating our explicit knowledge of probabilities with our implicit reading of patterns is a major challenge for traders.


Dr. Mezmer said...

In 1935, the psychologist and learning theorist Neal Miller conducted the following experiment.

To human subjects he presented in unpredictable order the symbols T (followed by electric shock) and 4 (not followed by shock). The shock was followed by a large galvanic response (GSR) that was soon conditioned not only by seeing the symbol T, but by anticipating it. From this and subsequent experiments Miller concluded that organisms should "behave 'foresightfully' because fear (i.e. anxiety), would be mediated by cues from a distinctive anticipatory goal response." Miller further concluded that the 'learned drive' of fear or anxiety, as marked by the GSR, obeys the same laws as do overt responses".

In other words anticipatory somatic changes mediate not choice, but avoidance, and may be described fully by learning principles.

Compare this experiment to the IGT experiment, where an individual again is confronted with a succession of symbols (in this case, markings on a card), and with unpredictable aversive consequences, in this case large negative card values occurring from time to time. If it assumed that unexpected 'bad information' is painful as well, then both experiments assume equivalence.

The difference however is that Miller assumed that tension based arousal or anxiety mediated avoidance, and Damasio's hypothesis assumed arousal mediated choice. So which one is right, and which one original? Those questions are as of yet, undecided.


Miller, N. E. (1971) Selected Papers, Atherton, Chicago



aj marr

Brett Steenbarger, Ph.D. said...

Hi AJ,

Great background on the studies and fascinating issues. It certainly makes evolutionary sense for somatic changes to mediate avoidance, but I wonder if there is a difference between the more nuanced, subtle markers that would guide preference and those that would lead to avoidance of aversive stimuli. Thanks for the excellent food for thought--


Dr. Mezmer said...

Good point.

In modern neurological accounts of incentive motivation, the release of the neuromodulator dopamine, which is felt as positive affect, also acts to somatically mark the goodness and badness of momentary decisions (think of the AHA moment when you have a great idea), but dopamine is measured by the MRI, not the SCR, and has no causal relationship with tension or autonomic arousal. The best source for knowledge on this is the neuropsychologist Kent Berridge's web site, which is a minor education in itself!


Brett Steenbarger, Ph.D. said...

Thanks, AJ, for the excellent observations and the reference to the web site. I look forward to staying in touch--


a said...

Great piece Dr Brett

I wonder what part corticosteroids play in the part if tripping up the normal pre cortex region mimicking that of the damaged control group?

My understanding is stress releases there hormones and tied in with the emotional lack of insight into these causes that making decisions under this situation seems to relate to thought processes being disrupted- in fact doing actual damage to the brain in those moments when integration with the trader's mind through his emotional sensory receptors and what is really going on on the trading floor opposite others with similar influences as well.

Your feedback controls are brilliant and my guess is they should determine a trading system specialized for that trader's emotional responses to risk instead of trying to adapt him to risk levels beyond his capacity to accept.