Q2. Discuss the neural mechanisms that underlie value-based decision making.
Consider a situation where a choice needs to be made between hunting for food and seeking a warm shelter. To decide between these two fundamentally different rewards, the brain needs to calculate the values and costs associated with each option, consider different motivational, cognitive and contextual variables, construct a plan to obtain reward outcomes, and finally input these signals into a choice process. This essay aims to consider the function of the different brain areas involved in the above processes and how they may work together in a network to give rise to value-based decision making.
Valuation-based learning and decision-making occur largely in the fronto-striatal network, involving four main regions, each responsible for a different function, as identified by Rushworth et al (2011): the ventromedial prefrontal cortex and adjacent medial orbitofrontal cortex (vmPFC/mOFC), lateral orbitofrontal cortex (lOFC), Anterior Lateral Prefrontal Cortex (aPFC) or Frontal Pole, and the Anterior Cingulate Cortex (ACC). Figure 1 depicts the above brain areas on a macaque brain, from which more precise research have been conducted in the field. While research in both human and macaque brains will be considered in this essay, neural imaging research in human often faces limitations such as the difficult-to-reach anatomical positions of these areas (many of them are located behind the eye lobes), and these brain regions being part of a “default network” (Raichle and Snyder, 2007), which means that they remain active at rest, and BOLD signals correspond to different degrees of deactivation compared to rest, rather than activation.
Figure 1. Frontal Brain Regions in the Macaque involved in valuation-based learning and decision-making. The vmPFC/mOFC is the most well documented area in the four regions mentioned above. BOLD signals in this area has consistently been shown to correlate with the reward value of a choice (Rushworth et al, 2011). Padoa-Schioppa & Assad (2006) measured firing rates of vmPFC/mOFC neurons when macaques decide between different volumes of either juice or water. While they usually prefer one drop of juice over one drop of water, larger volumes, e.g. four drops, could persuade them to choose water over the one drop of juice. Firing rates of these neurons were more likely to vary systematically with the value of the drinks, rather than with the drink’s physical properties, i.e. taste or volume. Wunderlich et al (2010) expanded these findings by showing that vmPFC/mOFC neurons decrease their responses to a food or water to zero when the reward values of food and water are reduced to zero by feeding to satiety, and these signals quickly reverses when the rewarding stimulus turns into a punisher. In humans, BOLD activation in this area correlates to the self-reported reward value of sensory stimuli such as taste, olfactory, somatosensory, visual, but also to abstract stimuli such as social approval and virtual monetary stimuli (Rolls, 2005). The above studies, among many other replications, provide compelling evidence for the role of the vmPFC/mOFC in valuation of a choice. They also suggest that value representations of different rewards are converted and compared on a common scale. Comparing rewards of different physical and subjective properties on a common scale may be an adaptive solution to deciding between a large number of options (Wallis et al, 2007). For instance, when assigning value to a new food type, an animal can gauge the value of this option relative to all other foods, rather than having to go through iterative comparisons between food type A versus B, B versus C, A versus C and so on.
It was suggested that vmPFC/mOFC encodes not only the expected benefit of the course of action but also the opportunity costs associated with the unchosen option. This stems from...