Trends in Cognitive Sciences
OpinionReason, emotion and decision-making: risk and reward computation with feeling
Introduction
Most of the decisions we make everyday involve uncertainty. Out at dinner, for example, a new menu item can sound tempting, but might be disappointing. An alternate route around a traffic jam might save time, but might make you even more late. A hot stock tip might strike it rich, but you might lose your entire investment. What processes underlie these decisions? A long tradition of research in judgment and decision making (JDM), stemming from choice or preference theory in microeconomics [1] and decision theory in philosophy [2], suppose that uncertain decisions are based on cognitive processes typically regarded as involving means-end reasoning, logical inference, mental effort and exact computation according to a cost-benefit calculus (Box 1). In the 1990s, however, JDM models increasingly incorporated emotional processes 3, 4, 5, influenced by a reconsideration of emotion in neuroscience [6]. As these models developed, a prevalent emphasis of emotional contributions to JDM was as approximate, heuristic processes that deliver rapid evaluations without mental effort 3, 5, 7, 8, 9, 10. In addition, JDM researchers increasingly accounted for conflict in decision making as the divergence between cognitive and emotional evaluations 3, 8, and pathological decision making as the result of affect heuristics 3, 5 (Box 2).
Despite the popularity and commonsense appeal of distinguishing between cognitive and emotional contributions to JDM, many fundamental issues remain unresolved. Theories can be characterized in terms of the representations and the computations over those representations they posit, and it remains unclear in what ways cognitive and emotional contributions to JDM differ along these dimensions. That is, at the level of representation, what specific parameters of uncertain decision contexts are encoded by the brain, to what extent do such representations correspond to the parameters of various decision-making frameworks, and to what extent do putatively distinct cognitive and emotional contributions to JDM correspond to distinct underlying representations of uncertain decision contexts? Addressing these issues poses several challenges, not least that competing theories are not behaviorally distinguishable. This suggests that adjudicating among different theories requires neural studies that use quantitative and parametric frameworks with suitable resolution to distinguish among the main parameters of these various models and disassociating the representation of their basic parameters from other potential components of uncertain choice, including learning, motivation and salience (see Ref. [11] for discussion). Based on recent work using such experimental designs, I suggest that putative distinctions between cognitive and emotional contributions to JDM at the level of representation collapse. In particular, I focus on emerging evidence suggesting that emotional contributions to JDM do not encode approximate, heuristic evaluations. Rather, it suggests that emotional processes encode the precise, mathematically defined parameters of traditionally cognitive accounts of decision-making from economics and related fields, such as finance. On a more general note, such findings indicate that once-considered basic distinctions, such as that between cognition and emotion, do not map seamlessly onto brain functioning. That is, just as studies of the deep interconnectivity among emotional and cognitive structures suggests that assigning cognitive or emotional specialization to structures is deeply problematic [12], proposed functional distinctions, such as complexity differences between emotional and cognitive representations and computations, are likewise problematic.
Section snippets
The minimal parameters of decision-making under uncertainty
To begin, it is necessary to identify the underlying representational schemes, or minimal parameters, that various JDM theories posit. Theories of decision-making under uncertainty rest on two fundamental representations of a decision context. Informally, a decision maker must represent both an estimate of the predicted value of a prospect (expected value) and how far away from the actual value that estimate might be (risk). This latter representation is essential to capture a pervasive feature
Evidence for neural correlates of JDM parameters
Numerous physiological studies have investigated brain responses to uncertainty in non-human primates and functional imaging in human brain responses to uncertainty (reviewed in Refs. 17, 18). To date, however, their emphasis has been on reward-related learning or choice rather than on the more basic question of underlying representations independent of learning, motivation and salience (see Ref. [18] for discussion). Among those examining this latter issue, there is some evidence for a
Reevaluating affective heuristic processing?
These results suggest several potential challenges for the affect heuristics view of emotional contributions to uncertain decisions, including neurobiological versions in terms of somatic markers (Box 3). On the one hand, reward and risk seem to be encoded in many of the brain areas that have been closely identified with emotional processing, including midbrain dopamine areas and insula. Midbrain dopamine areas are centrally implicated in both positive and negative emotion [25], motivated
Conclusions
The investigation of the neural basis of uncertain choice has progressed rapidly in the last few years, progressing from investigating neural responses to basic contrasts such as certainty versus uncertainty, to quantitative parametric frameworks capable of testing the extent to which brain activation reflects the parameters of formal JDM models. Perhaps the most surprising finding to date is that core emotional structures, including the midbrain dopamine system and insula, decompose uncertain
Acknowledgements
I would like to thank James Woodward and the anonymous reviewers for their insightful feedback on earlier versions of the manuscript. I am indebted to Kerstin Preuschoff and Peter Bossaerts for collaborations resulting in findings regarding risk and reward representation. The author's work is partially funded by the Gordon and Betty Moore Foundation.
References (55)
The neural bases of cognitive conflict and control in moral judgment
Neuron
(2004)Neural differentiation of expected reward and risk in human subcortical structures
Neuron
(2006)Investigating signal integration with canonical correlation analysis of fMRI brain activation data
Neuroimage
(2008)- et al.
The somatic marker hypothesis: a neural theory of economic decision
Games Econ. Behav.
(2005) Neural activity in the human brain relating to uncertainty and arousal during anticipation
Neuron
(2001)Increased activation in the right insula during risk-taking decision making is related to harm avoidance and neuroticism
Neuroimage
(2003)- et al.
The neural basis of financial risk taking
Neuron
(2005) The role of emotion in decision-making: evidence from neurological patients with orbitofrontal damage
Brain Cogn.
(2004)Neural signatures of economic preferences for risk and ambiguity
Neuron
(2006)- et al.
The hidden island of addiction: the insula
Trends Neurosci.
(2009)