Fig 1.
Two potential gambles for water reward were presented each trial. Gambles were represented by a rectangle, some proportion of which was grey, blue, or green, corresponding to a small, medium, or large reward respectively. The size of the grey, blue, or green portion corresponded to the probability that selecting that gamble would lead to the corresponding reward. Offers appeared in a random order one at a time with a one-second offset for 400 ms each. After fixation, both offers reappeared during a decision phase. Rewarded outcomes were accompanied by a white circle in the center of the chosen offer. These data are available in Data S1 on figshare (http://figshare.com/articles/Data_for_Signatures_of_value_comparison_in_ventral_striatum_neurons_/1332487).
Fig 2.
Coding of offer values in VS neurons.
A. Magnetic resonance image of monkey B. Recordings were made in the nucleus accumbens region of VS (highlighted in red; see S4 Fig for precise demarcation). B. Average responses of an example neuron (+/- 1 SEM in firing rate), separated by offer 1 reward size and probability. During epoch 1, this neuron showed higher firing rates for offers with larger reward sizes and probabilities. C. Scatter plot of each neuron’s coefficients for tuning for gamble probability (x-axis) and gamble reward size (y-axis). These coefficients were significantly correlated, consistent with a single value scale coding scheme. A least-squares regression line and confidence intervals are shown in red. Neurons are shown color-coded by regression coefficient p-value at α = 0.05. These data are available in Data S1 on figshare (http://figshare.com/articles/Data_for_Signatures_of_value_comparison_in_ventral_striatum_neurons_/1332487).
Table 1.
Neurons encode offer value in an abstract format.
Fig 3.
VS neuron activity related to comparison and choice.
A. Average responses of an example neuron (+/- 1 SEM in firing rate), separated by binned (lowest 33.3%, middle 33.3%, highest 33.3%) expected value difference between offer values (offer value 1 minus offer value 2). During the second offer presentation, this neuron fired more when offer value 2 was greater than offer value 1 (red), and less when offer value 1 was greater than offer value 2 (blue). B. Plot of the proportion of neurons that show a significant correlation between neural firing rate and the value of the 1st (blue) and 2nd (red) offers (500 ms sliding boxcar). Horizontal lines show when the proportion of cells shown reaches significance (dark gray: 5%; light gray: binomial test at α = .05). C. Scatter plot of each neuron’s coefficients for tuning for offer value 1 (x-axis) and for offer value 2 (y-axis), both in epoch 2. Least-squares regression line and confidence intervals are shown in red. Neurons are shown color-coded by regression coefficient p-value at α = 0.05. D. Scatter plot of each neuron’s coefficients for tuning for offer value 1 in epoch 1 (x-axis) and for offer value 2 in epoch 2 (y-axis). Neurons are shown color-coded by regression coefficient p-value at α = 0.05. These data are available in Data S1 on figshare (http://figshare.com/articles/Data_for_Signatures_of_value_comparison_in_ventral_striatum_neurons_/1332487).
Fig 4.
VS neurons come to code chosen value.
A. Average responses of an example neuron (+/- 1 SEM in firing rate), separated by binned expected value of the chosen offer. This neuron showed higher firing rates when the value of the eventually chosen offer was greater than average (blue), and lower firing when the chosen offer value was lower than average (red) starting in epoch 1 (knowing offer 1 gives the monkey partial information as to his eventual chosen offer) and extending into epochs 2 and 3. B. Plot of proportion of neurons that show a significant correlation between firing rates and the value of the chosen (VS: dark blue line; vmPFC: light blue line) and unchosen (VS: dark red line; vmPFC: light red line) offers (500 ms sliding boxcar). Note that each point on a sliding boxcar plot is derived from a 500 ms bin beginning at that point, so it is feasible for significant encoding to arise before offer presentation on the plot. Horizontal lines show when the proportion of cells shown reaches significance (dark gray: 5%; light gray: binomial test at α = .05 on VS dataset). vmPFC data from [15]. C. Proportion of cells whose firing rates significantly encoded value difference (chosen value-unchosen value). The horizontal lines show when the proportion of cells shown reaches significance (gray: 5%; purple: binomial test at α = .05 on VS dataset; green: binomial test at α = .05 on vmPFC dataset). D. Decodability of chosen offer in a 500 ms sliding boxcar for VS neurons (purple) and vmPFC neurons (green; see Methods). The horizontal lines show when the proportion of trials correctly classified reach significance (gray: 50%; purple: binomial test at α = .05 on VS dataset; green: binomial test at α = .05 on vmPFC dataset). These data are available in Data S1 (VS) and Data S2 on figshare (vmPFC; http://figshare.com/articles/Data_for_Signatures_of_value_comparison_in_ventral_striatum_neurons_/1332487).
Fig 5.
Coding of outcomes in VS neurons.
A. Average responses (+/- 1 SEM in spks/s) of an example neuron to task events, separated by gamble outcome. This neuron showed a negative tuning for outcome during epoch 3 (shaded area). B. Plot of proportion of neurons significantly tuned for gamble outcomes over time using a 500 ms sliding window. C. Same data as in B, but with tuning for outcome on previous trial instead of on current trial. Influence of the previous trial’s outcome was strong and lasted throughout the current trial. These data are available in Data S1 on figshare (http://figshare.com/articles/Data_for_Signatures_of_value_comparison_in_ventral_striatum_neurons_/1332487).