Dopamine, Salience, and Response Set Shifting in Prefrontal Cortex

Abstract

Dopamine is implicated in multiple functions, including motor execution, action learning for hedonically salient outcomes, maintenance, and switching of behavioral response set. Here, we used a novel within-subject psychopharmacological and combined functional neuroimaging paradigm, investigating the interaction between hedonic salience, dopamine, and response set shifting, distinct from effects on action learning or motor execution. We asked whether behavioral performance in response set shifting depends on the hedonic salience of reversal cues, by presenting these as null (neutral) or salient (monetary loss) outcomes. We observed marked effects of reversal cue salience on set-switching, with more efficient reversals following salient loss outcomes. L-Dopa degraded this discrimination, leading to inappropriate perseveration. Generic activation in thalamus, insula, and striatum preceded response set switches, with an opposite pattern in ventromedial prefrontal cortex (vmPFC). However, the behavioral effect of hedonic salience was reflected in differential vmPFC deactivation following salient relative to null reversal cues. l-Dopa reversed this pattern in vmPFC, suggesting that its behavioral effects are due to disruption of the stability and switching of firing patterns in prefrontal cortex. Our findings provide a potential neurobiological explanation for paradoxical phenomena, including maintenance of behavioral set despite negative outcomes, seen in impulse control disorders in Parkinson's disease.

Keywords: dopamine; fMRI; prefrontal cortex; reversal learning; set shifting.

Figure 1.

Task depiction. (A) Schematic of a reversal trial signaled by a salient monetary loss. Subjects have learnt the appropriate response to presentation of one of 2 paired symbols (Hiragana figures), either gripping (Go) or omitting a grip response (No-Go). Initially (left—trial before reversal, T_{reversal − 1}) subjects observe a symbol signaling a grip is required; a yellow border then appears to indicate that the grip has been registered successfully. Subsequently, an outcome screen appears. This example is taken from a block where subjects avoid a monetary loss when performing correctly; hence, an empty circle is shown to inform that no money was lost. On the next trial (middle—reversal trial, T_reversal), the contingencies have reversed and the elicited response of a grip is now incorrect, and the outcome screen signals loss of money (indicated by a pound sign with a cross through it). The alternative symbol (not shown) simultaneously switches its contingency from the complimentary “no-grip to avoid loss” to a “grip to avoid loss.” On the next trial (right—trial after reversal), the subject alters their behavioral response and now does not grip in response to the presentation of this symbol (indicated by the blue border). This is now the new correct response, and no money is lost (hence, the outcome screen shows an empty circle). (B) Schematic of a reversal trial signaled by a null outcome. In this example, subjects have learnt that the initial Hiragana symbol (left) signals that a grip is required, and a yellow border then appears to indicate that a grip has been registered successfully. In this block, the outcome screen indicates that they will receive money for a successful response (indicated by a pound coin). On the next trial (middle), the subject grips again; however, the contingencies have reversed and the elicited response of a grip is now incorrect. The outcome screen therefore signals a null outcome (indicated by an empty circle). On the next trial (right), the subject has changed behavior and now does not grip on observing the same symbol (indicated by the blue border) and hence receives money again for a successful response. In parallel, the alternative Hiragana symbol (not shown) will also switch contingencies on the reversal trial.

Figure 2.

Accuracy trials after reversal shift. Accuracy (percentage correct responses) in the trial after a reversal shift for both reversal types (salient reversal cue—gray bars/null reversal cue outcome—white bars), and drug conditions (placebo and l-dopa). Bars show within-subject standard error. Under placebo, subjects are more accurate when reversing their behavior after a salient reversal cue compared with a neutral cue. When subjects were given l-dopa, this performance advantage was eliminated (*P < 0.05; see text for statistical reporting).

Figure 3.

Neurophysiological responses to reversals in vmPFC. (A) Contrast of BOLD activation in the trial before a reversal (T_{reversal − 1}, fully expected outcome) minus the reversal trial (T_reversal, unexpected outcome) averaged over all conditions (placebo/l-dopa and salient/null). Trial-specific activations were modeled as box-car functions from trial onset, with durations set according to the trial length—as estimated on an individual subject basis. There was a significant reduction in vmPFC activation in response to a violation of expectations in reversal trials [Montreal Neurological Institute (MNI) space; peak (x, y, z) coordinates: −3, 11, −2; z = 3.84, P < 0.005 FWE-corrected] (B) Contrast estimates plotted for the reversal trial and the trial before reversal in the vmPFC (for trials with subsequent successful reversal shifts). There are differential neural responses in trials with expected and unexpected outcomes for the distinct trial types (salient loss trials—left; neutral/null trials—right). Avoiding losses under placebo (gray bars) leads to a positive BOLD response in vmPFC in the fully predicted trials and decreased BOLD response in reversal trials signaled by unexpected losses (*P < 0.05; for further statistical reporting see text). This pattern is not seen when individuals reverse their behavior following neutral cues. In contrast, under l-dopa (white bars) we observe the exact opposite pattern, with increased activation for expected outcomes and decreased activation following unexpected outcomes, paradoxically present in null trials and not (as is found under placebo) in the salient trials. Represented are the mean parameter estimates (β values). Bars show within-subject standard error.

Figure 4.

Neurophysiological responses to reversals in caudate, insula, and thalamus. (A) Contrast of BOLD activation in a reversal trial (T_reversal, unexpected outcome) minus the trial before reversal (T_{reversal − 1}, fully expected outcome), averaged over all conditions (placebo/l-dopa and salient/null). Differential activation is seen in bilateral insula, thalamus, and caudate [peak (x, y, z) MNI coordinates (z scores)—insula: −27, 17, 7 (4.43) and 42, 20, 1 (5.12); thalamus: −6, −19, 7 (4.74) and 6, −22, 7 (4.08); caudate: 12, 5, 7 (4.37); P < 0.005 FWE-corrected]. (B–D) Plots of mean parameter estimates (β values) for the reversal trial and the trial before reversal in the insula (B), thalamus (C), and caudate (D), contingent upon subsequent successful reversal shifting (i.e., a behaviorally validated response). Here, we observed a prediction error-type response (reduced activation at the time an unexpected/surprising outcome was presented) across trial types [salient loss trials on left; neutral/null trials on right; placebo (gray bars), l-dopa (white bars)]. (T_{reversal − 1}− T_reversal) contrast effects per condition: bilateral insula – (salient loss)_placebo: t_(1,14) = −5.93, P < 0.001, (null)_placebo: t_(1,14) = −3.41, P = 0.004; (salient loss)_l-dopa: t_(1,14) = −2.86, P = 0.012; (null)_l-dopa: t_(1,14) = −5.03, P < 0.001. Bilateral thalamus—(salient loss)_placebo: t_(1,14) = −4.65, P < 0.001, (null)_placebo: t_(1,14) = −3.14, P = 0.007; (salient loss)_l-dopa: t_(1,14) = −1.9, P = 0.07; (null)_l-dopa: t_(1,14) = −3.11, P = 0.008. Left caudate—(null)_placebo: t_(1,14) = −1.807, P = 0.092; (salient loss)_l-dopa: t_(1,14) = −2.803, P = 0.014. Within-subject standard error bars are shown. *P < 0.05; ⁺P < 0.1.