Neural correlates of stimulus-response and response-outcome associations in dorsolateral versus dorsomedial striatum

Abstract

Considerable evidence suggests that there is functional heterogeneity in the control of behavior by the dorsal striatum. Dorsomedial striatum may support goal-directed behavior by representing associations between responses and outcomes (R-O associations). The dorsolateral striatum, in contrast, may support motor habits by encoding associations between stimuli and responses (S-R associations). To test whether neural correlates in striatum in fact conform to this pattern, we recorded single-units in dorsomedial and dorsolateral striatum of rats performing a task in which R-O contingencies were manipulated independently of S-R contingencies. Among response-selective neurons in both regions, activity was significantly modulated by the initial stimulus, providing evidence of S-R encoding. Similarly, response selectivity was significantly modulated by the associated outcome in both regions, providing evidence of R-O encoding. In both regions, this outcome-modulation did not seem to reflect the relative value of the expected outcome, but rather its specific identity. Finally, in both regions we found correlates of the available action-outcome contingencies reflected in the baseline activity of many neurons. These results suggest that differences in information content in these two regions may not determine the differential roles they play in controlling behavior, demonstrated in previous studies.

Keywords: decision making; dorsal striatum; electrophysiology; goal-directed behavior; habit; learning; rat; single-unit activity.

Figure 1

Task structure, recording sites and behavioral data. (A) Shown are the sequences of events for trials during delay and size blocks. Each session consisted of four blocks, two delay blocks and then two size blocks. The short and long outcomes were randomly assigned to left or right on the first block, and on each subsequent block, the preferred outcome alternated sides. (B) Boxes show the estimated dorsal/ventral and medial/lateral extent of recording sites, based on the final position of the electrode. The range of the estimated rostral/caudal position, relative to bregma, is labeled on the figures. (C) Average choice rate, collapsed across direction, for the block transition from long delay to short delay or from small size to big size. The last 20 trials of the previous block are shown in gray shading. Note that each transition from long to short and small to big is accompanied by a transition, for the opposite response, in the opposite direction (i.e. from short to long and from big to small). Therefore, because choice rates for both responses must sum to 100%, the choice rates shown in these figures actually represent transitions in both directions. The bar graphs in the insets show average percent choice ± SEM of long vs. short delay or small vs. big size in the last 20 trials of blocks. Bar graphs in the lower panels show average reaction time (from odor offset to odor port exit) ± SEM on correct forced-choice trials on which the outcome was long vs. short delay, or small vs. big size, taken from the last 20 trials of all blocks. *p < 0.01 by t-test vs. opposite outcome.

Figure 2

Examples of free-choice/forced-choice-selective single-units recorded from dorsolateral and dorsomedial striatum. Shown are raster plots and time histograms displaying firing rate during correct forced-choice vs. free-choice trials, aligned on the beginning of movement for (A) and (C), and at the presentation of the odor for (B) and (D). (A) and (B) are units from dorsolateral striatum; (C) and (D) are units from dorsomedial striatum. All neurons shown here were significantly selective for the direction of movement that was executed on that trial, in these cases for the rightward movement. Other neurons (the minority) were selective for the leftward movement. Neurons shown in (A) and (D) were significantly selective for forced-choice trials, while those shown in (B) and (C) were significantly selective for free-choice trials. For forced-choice trials, only the last ∼80 trials are shown so that the raster plots are comparable in size to the free-choice raster plots.

Figure 3

Neurons selective for the upcoming response during the odor epoch in both dorsolateral (A) and dorsomedial (B) striatum show characteristics consistent with S–R encoding. Histograms show the free-choice/forced-choice index (average peak-normalized firing rate on free-choice trials minus that on matched forced-choice trials across all blocks) for all neurons that are selective for the upcoming response during the odor epoch (beginning 50 ms after presentation of the odor, ending at odor port exit). Those plotted in black are significantly selective for either forced-choice trials (negative values) or free-choice trials (positive values). Significance was tested using a paired t-test (p < 0.05) comparing free-choice trials with forced-choice trials matched for response, outcome, and position within the block. Asterisks in the histograms show the selectivity indices of the example neurons shown in Figures 2B,D. Curves on the right show average peak-normalized firing rates (±SEM), relative to baseline, aligned on the beginning of odor, for the free-choice selective population for dorsolateral striatum and the forced-choice selective population for dorsomedial striatum. Very few neurons in dorsolateral striatum were forced-choice preferring and very few in dorsomedial striatum were free-choice preferring. Curves were collapsed across each neuron's preferred direction (designated according to the direction and block with the highest average firing rate). Dorsomedial striatum included a significantly greater percentage of selective neurons, which were more likely to prefer forced-choice trials.

Figure 4

Neurons selective for the response during the movement epoch in both dorsolateral (A) and dorsomedial (B) striatum show characteristics consistent with S–R encoding. Histograms on the left show the free-choice/forced-choice index (average peak-normalized firing rate on free-choice trials minus that on matched forced-choice trials across all blocks) for all response-selective neurons. Those plotted in black are significantly selective for either forced-choice trials (negative values) or free-choice trials (positive values). Significance was tested using a paired t-test (p < 0.05) comparing free-choice trials with forced-choice trials matched for response, outcome, and position within the block. Asterisks in the histograms show the selectivity indices of the example neurons shown in Figures 2A,C. Curves on the right show average peak-normalized firing rates (±SEM), relative to baseline, aligned on the beginning of movement towards reward, for each significantly selective population. Curves were collapsed across each neuron's preferred direction (designated according to the direction and block with the highest average firing rate).

Figure 5

Examples of outcome-selective single-units recorded from dorsomedial and dorsolateral striatum. Shown are raster plots and time histograms displaying firing rate during correct forced-choice trials, aligned on the beginning of odor presentation in (A) and on the beginning of movement towards reward in (B). These two units, from dorsomedial (A) and dorsolateral (B) striatum, are both response-selective and show the greatest activity when a particular outcome is expected in their preferred direction. The unit in (A) responded most when the short-delayed outcome could be expected to result from the rightward movement, and the unit in (B) responded most when the long-delayed outcome could be expected to result from the rightward movement.

Figure 6

Neurons selective for the upcoming response during the odor epoch are modulated by outcome in both dorsomedial and dorsolateral striatum. Curves in (A) and (B) show average peak-normalized firing rates (±SEM), relative to baseline, during the last 20 forced-choice trials of each block, aligned on the beginning of odor presentation. Populations included all neurons selective for the upcoming response during the odor epoch (193 out of 587 recorded in dorsomedial; 147 out of 489 recorded in dorsolateral). Curves were collapsed across each neuron's preferred direction and preferred outcome (designated according to the direction and block with the highest average firing rate). Preferred outcomes were equally distributed across the four outcomes. Scatter plots in (C) and (D) show the delay modulation index vs. the size modulation index for each response-selective neuron. Colored points indicate neurons that were significantly selective for the size modulation, the delay modulation, or both. Bar graphs show the difference between the two indices for each neuron. To the extent that outcome-modulated responses reflect the value of the response, colored points should congregate around the diagonal, the colored bars should peak in the center, and the number of neurons significantly modulated by both manipulations should exceed chance. In fact, however, in both regions colored points are significantly removed from the diagonal and neurons modulated by both manipulations are no more frequent than chance. Thus separate populations of neurons encode each response–outcome conjunction. Delay modulation index = absolute value of the difference between normalized firing rates during preferred directional response on delay block 1 and delay block 2. Size modulation index is the corresponding difference for size blocks.

Figure 7

Neurons selective for the response during the movement epoch are modulated by outcome in both dorsomedial and dorsolateral striatum. Curves in (A) and (B) show average peak-normalized firing rates (±SEM), relative to baseline, during the last 20 forced-choice trials of each block, aligned on the beginning of the movement towards reward. Populations included all response-selective neurons (269 out of 587 recorded in dorsomedial; 237 out of 489 recorded in dorsolateral). Curves were collapsed across each neuron's preferred direction and preferred outcome (designated according to the direction and block with the highest average firing rate). Preferred outcomes were equally distributed across the four outcomes. Scatter plots in (C) and (D) show the delay modulation index vs. the size modulation index for each response-selective neuron. Colored points indicate neurons that were significantly selective for the size modulation, the delay modulation, or both. Bar graphs show the difference between the two indices for each neuron. To the extent that outcome-modulated responses reflect the value of the response, colored points should congregate around the diagonal, the colored bars should peak in the center, and the number of neurons significantly modulated by both manipulations should exceed chance. In fact, however, in both regions colored points are significantly removed from the diagonal and neurons modulated by both manipulations are no more frequent than chance. Thus separate populations of neurons encode each response–outcome conjunction. Delay modulation index = absolute value of the difference between normalized firing rates during preferred directional response on delay block 1 and delay block 2. Size modulation index is the corresponding difference for size blocks.

Figure 8

Outcome-selectivity in both dorsomedial and dorsolateral striatum depends on the preceding stimulus. Curves in (A) and (B) show average peak-normalized firing rates (±SEM), relative to baseline, on matched forced-choice and free-choice trials, aligned on the beginning of the movement towards reward. Curves were collapsed across each neuron's preferred direction and preferred outcome (designated according to the direction and block with the highest average firing rate on forced-choice trials). Only the latter half of free-choice trials in each block, along with forced-choice trials matched for direction, outcome and position within the block, are included. Only response-selective neurons from sessions in which all conditions had at least two free-choice trials are included, resulting in 87 neurons in dorsomedial striatum and 122 neurons in dorsolateral striatum.

Figure 9

Examples of block-selective single-units recorded from dorsolateral and dorsomedial striatum. Shown are raster plots and time histograms displaying firing rate during correct forced-choice trials, aligned on the beginning of the trial. Each row includes the trials from one block of the session. The unit shown in (A), from dorsolateral striatum, shifted its baseline firing rate in the block with long-delayed outcomes on the left and short-delayed outcomes on the right. The unit shown in (B), from dorsomedial striatum, shifted its baseline firing rate in the block with big outcomes on the left and small outcomes on the right. Blocks are shown in the temporal order in which they occurred.

Figure 10

Baseline firing rates reflect available outcomes in both dorsolateral (A) and dorsomedial (B) striatum. Curves show average peak-normalized firing rates (±SEM) during the last 20 forced-choice trials of each block, aligned on the beginning of the trial, collapsed across each neuron's preferred block (designated according to the block with the highest average firing rate). Neurons in these populations showed an elevated baseline firing rate during their preferred block. Populations included all neurons with a significant effect of block, but no effect of direction. (99 out of 489 recorded in dorsolateral; 112 out of 587 recorded in dorsomedial).

Figure 11

Block-selective shifts of the baseline firing rate in dorsolateral (A) and dorsomedial (B) striatum develop across the block and return in the subsequent block. Curves show average peak-normalized firing rates (±SEM) during the pre-response epoch (from the beginning of the trial to the beginning of the response) across the preferred block, and, for comparison, across the block with the same-valued outcomes in the same directions. Populations are the same as those shown in Figure 10. The increase in baseline firing rate developed across the preferred block as the rat learned the response–outcome contingencies, returned to its original level during the following block, and did not change during other blocks. These changes in the baseline firing rate are more consistent with encoding outcomes that are available on a particular block than with recording artifacts. First blocks of sessions were excluded from this analysis.