Characterizing the associative content of brain structures involved in habitual and goal-directed actions in humans: a multivariate FMRI study

Abstract

While there is accumulating evidence for the existence of distinct neural systems supporting goal-directed and habitual action selection in the mammalian brain, much less is known about the nature of the information being processed in these different brain regions. Associative learning theory predicts that brain systems involved in habitual control, such as the dorsolateral striatum, should contain stimulus and response information only, but not outcome information, while regions involved in goal-directed action, such as ventromedial and dorsolateral prefrontal cortex and dorsomedial striatum, should be involved in processing information about outcomes as well as stimuli and responses. To test this prediction, human participants underwent fMRI while engaging in a binary choice task designed to enable the separate identification of these different representations with a multivariate classification analysis approach. Consistent with our predictions, the dorsolateral striatum contained information about responses but not outcomes at the time of an initial stimulus, while the regions implicated in goal-directed action selection contained information about both responses and outcomes. These findings suggest that differential contributions of these regions to habitual and goal-directed behavioral control may depend in part on basic differences in the type of information that these regions have access to at the time of decision making.

Keywords: MVPA; decision making; fMRI; goal-directed; habitual.

Figure 1.

Task design and time-span decoding. A, Subjects performed a binary decision task. One of two possible initial stimuli (STIMULUS) was presented, which determined the subsequent deterministic action-outcome contingencies between two possible actions and two possible outcome states. Outcome states were denoted by either a blue circle or a red square and were followed, after a short delay, by one of two distributions of monetary rewards (large or small). Crucially, each possible combination of stimulus, action, and outcome was permuted across conditions, thereby ensuring that representations of these different decision variables were not conflated (see Material and Methods). B, These analyses aimed to assess the classification of events that were being associatively retrieved by the initial stimulus. The classifier was trained on the relevant events (e.g., actions) at the time of their execution, but tested on those same events at the time of the initial stimulus (see Materials and Methods for more details).

Figure 2.

Behavior. A, Choice accuracy. B, Response times across training (Session 1) and experimental sessions. Each time bin represents four trials, and each data point is averaged across conditions and participants.

Figure 3.

Goal-directed representations. A, Right dlPFC encoded both action and outcome representations at the time of the initial stimulus presentation (conjunction analysis, x = 60, y = 17, z = 34, t₍₁₈₎ = 3.47). B, vmPFC encoded action at the time of initial stimulus presentation and outcome information at the time of action performance (conjunction, x = 3, y = 53, z = −20, t₍₁₈₎ = 5.54). C, Bar plot depicts accuracy score distributions in an independently defined dlPFC ROI. This score is the decoding accuracy minus 0.5, which is the expected accuracy of a random algorithm, *p < 0.05, **p < 0.005. D, Bar plot depicts accuracy score distributions for vmPFC, ***p < 0.0005.

Figure 4.

Habitual representations. A, A region of DLS (posterior putamen), extending into the globus pallidus (GP) was found to encode information about the action to be performed at the time of initial stimulus presentation (p < 0.05 SVFWE, x = −27, y = −22, z = 7, t₍₁₈₎ = 3.24), but critically, no significant information about outcome. B, The distribution of accuracy scores for actions and outcomes at the time of initial stimulus in an independently defined putamen/GP region of interest, ***p < 0.0005.

Figure 5.

Integrated stimulus-action representations were localized in DMS (anterior caudate nucleus; p < 0.05 SVFWE, x = 15, y = 11, z = 22, t₍₁₈₎ = 4.16) and hippocampus (p < 0.05 SVFWE, x = 24, y = −1, z = −20, t₍₁₈₎ = 4.35).