Taste-guided decisions differentially engage neuronal ensembles across gustatory cortices

Abstract

Much remains to be understood about the differential contributions from primary and secondary sensory cortices to sensory-guided decision making. To address this issue we simultaneously recorded activity from neuronal ensembles in primary [gustatory cortex GC)] and secondary gustatory [orbitofrontal cortex (OFC)] cortices while rats made a taste-guided decision between two response alternatives. We found that before animals commenced a response guided by a tastant cue, GC ensembles contained more information than OFC about the response alternative about to be selected. Thereafter, while the animal's response was underway, the response-selective information in ensembles from both regions increased, albeit to a greater degree in OFC. In GC, this increase depends on a representation of the taste cue guiding the animal's response. The increase in the OFC also depends on the taste cue guiding and other features of the response such as its spatiomotor properties and the behavioral context under which it is executed. Each of these latter features is encoded by different ensembles of OFC neurons that are recruited at specific times throughout the response selection process. These results indicate that during a taste-guided decision task both primary and secondary gustatory cortices dynamically encode different types of information.

Figure 1.

Two-alternative forced choice task. a, An illustration of a trial. Licks on a particular spout are depicted as vertical ticks. Color-coded bars are placed over licks that are followed by solution delivery (red = NaCl, blue = water). RT and MT are indicated by lines that mark their beginning and end and are aligned to the relevant trial events. The Epochs E0–E3 and Baseline (B), each being 500 ms, are highlighted in beige during the trial, and described in the text. The delivery of a particular cue (i.e., a single NaCl concentration) determines which choice response will most likely result in reward. b, This plot depicts the mean probability [P(“high”)], of the rat choosing the High choice spout as a function of NaCl concentration. c, A bar graph showing the relative proportions of event-related neurons from E0–E3 in GC (red) and OFC (blue). A black line that bridges two bars indicates a significant difference (p < 0.05; χ² test) between the groups represented by each bar.

Figure 2.

GC and OFC neurons show selective responses during trial-relevant Epochs. The raster plots and PSTHs of representative neurons from five neuron classes. A representative GC and OFC neuron from each class is shown with reference to the Baseline (B) and Epochs E0–E3 as seen in Figure 1a and described in the text. The regions are separated by double black lines. Within each region, a neuron class is represented by a row and each column denotes the trial Epoch (B and E0–E3) to which the PSTH is time-locked. The PSTHs reflect all trials in the session, regardless of response selection.

Figure 3.

Event-related neurons in GC and OFC. a, Population PSTH (mean ± 99% confidence interval) for E0_EX-neurons from GC (red) and OFC (blue) referenced to the start of E0. b, Population PSTH for E0_IN-neurons from GC (red) and OFC (blue) referenced to the start of E0. c, Population PSTH for E1 neurons from GC (red) and OFC (blue) referenced to the end of E1 and beginning of E2. d, Population PSTH for E2 neurons from GC (red) and OFC (blue) referenced to the end of E1 and beginning of E2. e, Population PSTH for E3 neurons from GC (red) and OFC (blue) referenced to the end of E3. In all panels data are presented as mean ± 99% confidence interval.

Figure 4.

The ensemble discrimination index (eDI) reflects a neuronal population's response selectivity. a, Two PSTHs (20 ms time bins) and associated raster plots are shown for a single OFC neuron under the Low response (left) and High response (right) selection condition. The PSTHs are referenced to the end of E3, which is highlighted in beige. The PSTH during E3 in each response selection condition is converted into a column vector (activity vectors) reexpressing activity as a normalized value between 0 and 1. Each column vector is directly compared using the equation depicted to generate an eDI value = 0.098. b, The same process as described above is performed for a different OFC neuron with an eDI value = 0.019, indicating the response is more similar to that than the neuron's response depicted in a.

Figure 5.

The ensemble discrimination index (eDI) increases from E0 to E3. a, The eDI (mean ± SEM) for E0–E3 as a function of region. b, The eDI for E0–E3 for data in a displayed as a line-plot to better illustrate the eDI increase across Epochs and regions.

Figure 6.

The laterally directed choice (E1–E3) responses are compared with the laterally directed responses that precede the start of a trial (C1–C3). a, A continuation of the single trial depicted in Figure 1a, in relation to the rat's movements after a water reward is delivered. Licks on a particular spout are depicted as vertical ticks. Color-coded bars are placed over licks that are followed by solution delivery (blue = water). The C1–C3 time periods are highlighted in beige together with the E1–E3 time periods. Black lines that connect the E1–C1, E2–C2 and E3–C3 time periods are considered to be analogous in regard to the type of behavior in which the rat is engaging. b, Depicts the left- and right-directed movement in egocentric space for the Post-Cue (left) and Post-Choice (right) condition.

Figure 7.

The increase in the ensemble discrimination index (eDI) from E1–E3 depends on taste guiding response selection in GC but to a lesser extent in OFC. a, The mean eDI for GC (mean ± SEM) across analogous Epochs (E1–C1, E2–C2 and E3–C3) and as a function of Post-Cue or Post-Choice condition. b, The eDI (mean ± SEM) in GC for E0–E3 as a function of Post-Cue or Post-Choice displayed as a line-plot. c, The mean eDI for OFC (mean ± SEM) across analogous Epochs (E1–C1, E2–C2 and E3–C3) and as a function of Post-Cue or Post-Choice condition. d, The mean eDI (mean ± SEM) in OFC for E0–E3 as a function of Post-Cue or Post-Choice displayed as a line-plot.

Figure 8.

The OFC contains three distinct neuronal ensembles that are differentiated by when they become active in relation to the Choice response. a, A bar graph showing the relative proportions of neurons active in the Post-Cue (left), Post-Choice (center), or Both (right) conditions in GC (red) and OFC (blue). A black line that bridges two bars indicates a significant difference (p < 0.05; χ² test) between the groups represented by each bar. b, The eDI (mean ± SEM) for OFC Both neurons during the Post-Cue (light blue) or Post-Choice (dark blue) condition and as a function of Epoch (where we assumed E1C1, E2C2 and E3C3 for the analysis). c, The eDI (mean ± SEM) for OFC Both neurons (light blue—see text) and OFC Post-Choice neurons (black—see text) only during the Post-Choice condition and expressed as a function of C1–C3.