Dissociable regulation of instrumental action within mouse prefrontal cortex

Abstract

Evaluation of the behavioral 'costs', such as effort expenditure relative to the benefits of obtaining reward, is a major determinant of goal-directed action. Neuroimaging evidence suggests that the human medial orbitofrontal cortex (mOFC) is involved in this calculation and thereby guides goal-directed and choice behavior, but this region's functional significance in rodents is unknown despite extensive work characterizing the role of the lateral OFC in cue-related response inhibition processes. We first tested mice with mOFC lesions in an instrumental reversal task lacking discrete cues signaling reinforcement; here, animals were required to shift responding based on the location of the reinforced aperture within the chamber. Mice with mOFC lesions acquired the reversal but failed to inhibit responding on the previously reinforced aperture, while mice with prelimbic prefrontal cortex lesions were unaffected. When tested on a progressive ratio schedule of reinforcement, mice with prelimbic cortical lesions were unable to maintain responding, resulting in declining response levels. Mice with mOFC lesions, by contrast, escalated responding. Neither lesion affected sensitivity to satiety-specific outcome devaluation or non-reinforcement (i.e. extinction), and neither had effects when placed after animals were trained on a progressive ratio response schedule. Lesions of the ventral hippocampus, which projects to the mOFC, resulted in similar response patterns, while lateral OFC and dorsal hippocampus lesions resulted in response acquisition, though not inhibition, deficits in an instrumental reversal. Our findings thus selectively implicate the rodent mOFC in braking reinforced goal-directed action when reinforcement requires the acquisition of novel response contingencies.

Fig. 1. Distinct response patterns in an instrumental reversal task in mice with discrete medial prefrontal lesions

(a) Experimental timeline. Mice were first trained to perform an instrumental response for food. The break in the timeline corresponds to surgery and recovery. Next, mice were given 2–3 “reminder” sessions, then the response requirement was reversed on the first of several sessions that constitute the response acquisition and suppression curves below. (b) Mice with mOFC lesions appropriately shifted instrumental responding to the newly reinforced aperture, (c) but were unable to coincidentally suppress “perseverative” responding on the previously reinforced aperture. (d) Responding on the newly reinforced aperture during reversal was unaffected by PL lesions. (e) Perseverative responding was also unchanged. (f) Composites of the largest and smallest lesions for all mOFC experiments are shown. (g) “PL” lesions are also represented; note that approximately one-quarter of these lesions included the infralimbic cortex, as here. All lesions were bilateral, and atlas images are reprinted with permission from Paxinos and Franklin (2003) with coordinates relative to bregma indicated. Symbols represent means (+SEM) per treatment group (*p<0.05 compared to sham).

Fig. 2. Medial prefrontal lesions regulate progressive ratio break point ratios

(a) When tested on a progressive ratio schedule of reinforcement, break point ratios achieved by mice with mOFC lesions were initially indistinguishable from sham levels on session 1, but then escalated. By contrast, break point ratios in mice with PL lesions appeared to decline. (b) To verify this impression, break point ratios on days 2–5 were normalized to those achieved on day 1, revealing consistently lower response levels in PL mice. (c) Representative GFAP staining for mice in this experiment is shown with mOFC lesion at top and PL at bottom. Symbols represent means (+SEM) per treatment group (*p<0.05 compared to sham).

Fig. 3. Sensitivity to outcome devaluation, non-reinforcement, and progressive ratio training prior to lesion placement

(a) Post-training medial PFC lesions did not affect sensitivity to satiety-specific outcome devaluation, as indicated by reduced responding in all groups after the devaluation procedure relative to responses made in the absence of devaluation (represented by the dashed line at 100%). (b) Similarly, when reinforcement was withheld in extinction training, all mice showed the expected decline in responding with no differences between groups. (c) A separate group of mice was trained to respond on a progressive ratio schedule of reinforcement prior to surgery, then tested again after lesions were placed. Under these conditions, neither medial PFC lesion affected responding. Bars here represent mean break point ratios achieved during 5 test sessions prior to lesion placement and 5 sessions after (+SEM). Otherwise, symbols and bars represent means (+SEM) for individual sessions.

Fig. 4. vHC lesions mimic mOFC lesions

(a) Because the vHC projects to the mOFC, we generated mice with vHC lesions to compare response patterns. Again, acquisition of responding on the previously non-reinforced aperture was unaffected, while (b) “perseverative” responding in reversal was exaggerated. (c) vHC lesions also increased break point ratios. (d) Histological analyses verified GFAP staining predominantly in the vHC; gray represents the largest lesion and black the smallest. All lesions were bilateral, and coordinates relative to bregma are indicated. Symbols represent means (+SEM) per treatment group (*p<0.05 compared to sham).

Fig. 5. lOFC and dHC lesions produce distinct response patterns

(a) Mice with lesions of the lOFC or dHC were also tested. lOFC lesions delayed the acquisition of responding on the newly reinforced nose poke, resulting in significantly fewer responses during session 3, but not later. dHC lesions also retarded reversal, resulting in less responding during the final test sessions (6–7). (b) Mice with dHC lesions appropriately extinguished responding on the non-reinforced aperture, while lOFC mice decreased responding during session 1. (c) dHC lesions elevated break point ratios, while lOFC lesions had no significant effects. (d) GFAP staining confirmed lOFC lesions spared medial PFC structures; representative GFAP staining at ~bregma 2.8 is shown. (e) Lesion composites are also provided, with gray outlining the largest lesions and black the smallest. All lesions targeted the lOFC, most spread to the ventral OFC, and 28% of lOFC lesions also resulted in GFAP staining in the dorsolateral orbital cortex (“DLO” in Paxinos and Franklin, 2003) as indicated at top. (e) dHC lesions are also represented; note the vHC is largely spared. All lesions were bilateral, and coordinates relative to bregma are indicated. Symbols represent means (+SEM) per treatment group (*p<0.05, **p<0.001 compared to sham).