The Neurobiology of Decision

Notes
The identied components of primate decision making (which, when basic enough, is similar to human
decision making) are in the ventromedial prefrontal cortex and associated parts in the striatum
(multicomponent valuation stage); and in the lateral prefrontal and parietal areas (choice stage).
Therefore, in PDF we are only looking at the choice stage because of the temporal resolution
limitations that fMRI possesses. So in FEF AND parietal areas, whatever has a higher activation (attn
or sacc) should be more involved in the decision making process.
Any neural model of decision making has to answer two questions:
First, how are the subjective values of the various options under consideration learned, stored, and
represented?
Second, how is a single highly slued action chosen from among the options under consideration to be
implemented by the motor circuitry?
Neurophysiological models serve to nd the limits of behavior so that the theorists will be able to make
better algorithms and better predict behavior.
Common currency, i.e. the abstract value that represents the value of the things being decided over, is
controlled by a disparate population of neurons.
Offer value neurons judge the value of the object (aka offer) while chosen value neurons maintain that
abstract value of judgement.
Padoa-Schioppa and Assad (2008): Showed menu invariance, i.e. neuronal responses stay constant for
an option in a decision. For example, neuronal responses for grape juice neurons will be the same even
if you have the choice between either apple juice or orange juice.
Lau and Glimcher (2008): Recorded from phasic ally active stratal neurons (PANs) in the caudate
nucleus and found three kinds of task-related responses:
Action value neurons
Track the value of one of the actions, independent of whether it was chosen.
Chosen value neurons
Track the value of a chosen action
Choice neurons
Produce a categorical response when a particular action is taken.
To summarize, a neuron has a consistent representation for a choice. For example when choosing
between right or left, the neurons responsible for the left choice would re at an intermediate level if the
probability for reward after the left choice is 50%, regardless of if the right choice has a probability of
giving a reward of 90%.
Plassmann and colleagues (2007): Scanned hungry subjects bidding on various snack foods. They
found that BOLD activity in the medial orbitofrontal cortex was correlated with the subjects subjective
valuation of that item.
Montague et al. (1996): dopaminergic neurons encode a reward prediction error i.e. the dopamine
release depends on the fulllment of the subjects prediction. For example, if the person was expecting to
get shocked but did not get shocked, the dopaminergic neurons would respond less. If the person did get
shocked then the dopaminergic neurons would not re.
Glimcher and Sparks (1992): showed that if two saccadic targets of roughly equal subjective value were
presented to a monkey, then the two locations on this map corresponding to the two saccades became
weakly active. If one of these targets was suddenly identied as having higher value, this led almost
immediately to a high-frequency burst of activity at the site associated with that movement and a
concomitant suppression of activity at the other site.
Platt and Glimcher (1999): showed that desirability of an action was linearly correlated with higher ring
of the neurons in the LIP that preferred that direction. Therefore, since the LIP/fronto-parietal network
has a population of neurons for a specic direction and magnitude of a saccade, then the amount of
ring in a certain population is indicative of a higher desirability of making a saccade in that certain
direction and magnitude.
Firing rates in LIP are not menu invariant i.e. each neurons ring rate changes relative to the other
neurons ring rates.
Taosheng was right. What were looking at is all relative. ALL RELATIVE.