Outliers

How does the brain deal with outlier risk?

Extreme events, or “outlier” (to use a statistical term) come in several flavors. Psychology has traditionally focused on salient but infrequent outliers (“oddballs”, “freak events”) and humans are known to over-react to those. Computational neuroscience has traditionally studied outliers that signal a change in contingencies (either in reversal tasks or restless bandits) which call for renewed exploration and learning. Here, outliers are frequent, but not salient, so that the decision maker has to pay close attention. The neurobiology underlying the remarkable ability of humans to adapt correctly in such an environment has become elucidated. The evidence points to anterior cingulate cortex and the cholinergic/adrenergic brain system in tracking outliers and controlling responses.

Our own work has started to study human reaction to a third kind of extreme events, namely, outliers that are both salient and frequent. These tend not to occur in one’s natural environment, but mostly emerge as the result of large-scale social interaction such as in internet traffic, or, perhaps most importantly, financial markets. We discovered that humans have difficulty coping with this type of outlier. In ensuring fast, adaptive response, anterior insula emerged as a crucial mediator, thus clarifying (and explaining) our earlier finding that this brain region specializes in tracking random shifts in risk (“stochastic volatility”), the cause behind this unusual type of outlier.

Recent publications by members of the Brain, Mind & Markets Lab on this topic:
d’Acremont M, Bossaerts P. Human reaction to leptokurtic noise and its neurobiological foundations. Cerebral Cortex, 26:1818-30, 2016. [PDF]