The neurobiology of decision: consensus and controversy

Abstract

We review and synthesize recent neurophysiological studies of decision making in humans and nonhuman primates. From these studies, the basic outline of the neurobiological mechanism for primate choice is beginning to emerge. The identified mechanism is now known to include a multicomponent valuation stage, implemented in ventromedial prefrontal cortex and associated parts of striatum, and a choice stage, implemented in lateral prefrontal and parietal areas. Neurobiological studies of decision making are beginning to enhance our understanding of economic and social behavior as well as our understanding of significant health disorders where people's behavior plays a key role.

Figure 1

Valuation circuitry. Diagram of a macaque brain, highlighting in black the regions discussed as playing role in valuation. Other regions are labeled in grey.

Figure 2

An example orbitofrontal neuron that encodes offer value, in a menu-invariant and therefore transitive manner. (a) In red is the firing rate of the neuron (± s.e.m.), as a function of the magnitude of the two juices offered, for three different choice pairs. In black is the percentage of time the monkey chose the first offer. (b) Replots firing rates as a function of the offer value of juice C, demonstrating that this neuron encodes this value in a common currency in a manner that is independent of the other reward offered. The different symbols and colors refer to data from the three different juice pairs, and each symbol represents one trial type. Reprinted with permission from Padoa-Schioppa and Assad (2008).

Figure 3

Two example striatal neurons that encode action value. (a) Caudate neuron that fires more when a contralateral saccade is more valuable (blue) compared to less valuable (yellow), independently of which saccade the animal eventually chooses. c denotes the average onset of the saccade cue. Reprinted with permission from Lau and Glimcher (2008). (b) Putamen neuron that encodes the value of a rightward arm movement (Q_R), independent of the value of a leftward arm movement (Q_L). Reprinted with permission from Samejima et al. (2005).

Figure 4

Orbitofrontal cortex encodes the subjective value of food rewards in humans. (a) Hungry subjects bid on snack foods, which were the only items they could eat for 30 minutes after the experiment. At the time of the decision, medial orbitofrontal cortex (b) tracked the subjective value that subjects placed on each food item. Activity here increased as the subjects willingness-to-pay for the item increased (c). Reprinted with permission from Plassmann et al. (2007).

Figure 5

Match between psychometric and neurometric estimates of subjective value during intertemporal choice. (a) Regions-of-interest are shown for one subject, in striatum, medial prefrontal cortex and posterior cingulate cortex. (b) Activity in these ROIs (black) decreases as the delay to a reward increases, in a similar manner to the way that subjective value estimated behaviorally (red) decreases as a function of delay. This decline in value can be captured by estimating a discount rate (k). (c) Comparison between discount rates estimated separately from the behavioral and neural data across all subjects, showing that on average there is a psychometric-neurometric match. Reprinted with permission from Kable and Glimcher (2007).

Figure 6

Dopaminergic responses in monkeys and humans. (a) An example dopamine neuron recorded in a monkey, which responds more when the reward received was better than expected. (b) Firing rates of dopaminergic neurons track positive reward prediction errors. (c) Population average of dopaminergic responses (n=15) recorded in humans during deep brain stimulation (DBS) surgery for Parkinson’s disease, showing increased firing in response to unexpected gains. The red line indicates feedback onset. (d) Firing rates of dopaminergic neurons depend on the size and valence of the difference between the received and expected reward. All error bars represent standard errors. Panels (a–b) reprinted with permission from Bayer and Glimcher (2005), and panels (a–b) reprinted with permission from Zaghloul et al. (2009).

Figure 7

Choice circuitry for saccadic decision-making. Diagram of a macaque brain, highlighting in black the regions discussed as playing a role in choice. Other regions are labeled in grey.

Figure 8

LIP firing rates are greater when the larger magnitude reward is in the response field (n=30) (a), but are not affected when the magnitude of all rewards are doubled (n=22) (b). Adapted with permission from Dorris and Glimcher (2004).

Figure 9

Schematic of the symmetric random walk (a) and race models (b) of choice and reaction time. (c) Schematic neural architecture and simulations of the computational model of that replicates activation dynamics in LIP and Superior Colliculus during choice. Panels (a) and (b) reprinted with permission from Gold and Shadlen (2007). Panel (c) adapted with permission from Lo and Wang (2006).