Encoding and tracking of outcome-specific expectancy in the gustatory cortex of alert rats

Abstract

In natural conditions, gustatory stimuli are typically expected. Anticipatory and contextual cues provide information that allows animals to predict the availability and the identity of the substance to be ingested. Recording in alert rats trained to self-administer tastants following a go signal revealed that neurons in the primary gustatory cortex (GC) can respond to anticipatory cues. These experiments were optimized to demonstrate that even the most general form of expectation can activate neurons in GC, and did not provide indications on whether cues predicting different tastants could be encoded selectively by GC neurons. Here we recorded single-neuron activity in GC of rats engaged in a task where one auditory cue predicted sucrose, while another predicted quinine. We found that GC neurons respond differentially to the two cues. Cue-selective responses develop in parallel with learning. Comparison between cue and sucrose responses revealed that cues could trigger the activation of anticipatory representations. Additional experiments showed that an expectation of sucrose leads a subset of neurons to produce sucrose-like responses even when the tastant was omitted. Altogether, the data show that primary sensory cortices can encode for cues predicting different outcomes, and that specific expectations result in the activation of anticipatory representations.

Keywords: anticipatory cues; coding; expectation; insular cortex; learning; reward.

Figure 1.

Experimental design. A, Schematic representing the reconstructed positioning of recording electrodes in GC. Each line represents the positioning of a movable bundle along the dorsoventral axis. Circles, Starting point of recordings; crosses, final recording site. Positioning was reconstructed according to initial stereotaxic coordinates, electrode advancement after a group of sessions, and histological identification of the terminal recording site. S1, Somatosensory cortex; GI, granular insular cortex; DI, disgranular insular cortex; AID, dorsal agranular insular cortex; AIV, ventral agranular insular cortex; CL, claustrum; Pir, piriform cortex. B, Behavioral paradigm: auditory go/no-go task. Top, Design of each trial. Bottom, A representative session. x-Axis, trials; y-axis, percentage of cues followed by a lever press; cyan line, Suc trials; gold line, Q trials; black line, performance averaged across cues; dotted line, level of criterion performance. Performance was computed using a moving window of 10 trials and a step of 1 trial. C, Raster plots and PSTHs for a representative cue-responsive neuron during a complete session. Left panels, Raster plots and PSTHs in response to Suc_cue (top) and Q_cue (bottom). Vertical line at time 0 indicates the onset of the auditory cue. Right panels, Raster plots and PSTHs in response to self-delivery of Suc (top) and Q (bottom). Vertical line at time 0 indicates the time of lever pressing, self-delivery, and the simultaneous offset of the cue. Cyan triangle, Time of Suc self-delivery; gold triangle, time of Q self-delivery; cyan circle, time of Suc_cue onset; gold circle, time of Q_cue onset; red circle, ITI lever presses (i.e., extra presses not leading to self-delivery); dashed horizontal line, trial at which criterion is achieved. In the raster plots, the top rows represent earlier trials.

Figure 2.

Neurons in the GC of well trained rats respond to specific cues. A, Representative raster plots and PSTHs of a Suc_cue-selective neuron (cyan); a Q_cue-selective neuron (gold); a neuron responding to both cues, but with stronger responses to Q_cue (magenta); and a neuron responding equally to both cues (black). Top two panels, Responses to Suc_cue; bottom two panels, responses to Q_cue; triangles in the rasters, time at which the tastant was self-administered (cyan, Suc; gold, Q); red dashed line superimposed on the PSTHs, time course of mouth movements; x-axis, time (vertical line 0 indicates the onset of the cue); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; y-axis for mouth movements, difference of intensity for adjacent pixels. B, Population PSTH of excitatory and inhibitory responses averaged across cues. C, Absolute difference between responses to Suc_cue and Q_cue averaged across all cue-responsive neurons. The thick line above the trace represents a significant difference from baseline. For A–C: x-axis, time (0 is the onset of the cue); y-axis, auROC; shading around traces, SEM. Shaded vertical rectangles show the time window in which cue responsiveness was considered. D, Distribution of cue-responsive neurons as follows: Suc_cue-selective (cyan); Q_cue-selective (gold); differentially responding to both cues (magenta); and equally responding to both cues (black). Dark shades, Excitatory responses; light shades, inhibitory responses; intermediate shade of magenta, neurons with excitatory and inhibitory responses depending on the cue. E, Quantification of selectivity on the basis of ΔPSTH. Cyan/gold stripes, Suc_cue- and Q_cue-selective neurons; magenta, neurons differentially responding to both cues; black, neurons equally responding to both cues; gray bar, neurons not responsive to cue. The asterisk indicates significant differences across all combinations.

Figure 3.

Cue responses precede mouth movements. A, Population PSTH for a subset of Suc_cue-selective (left, cyan traces) and Q_cue-selective (right, gold traces) neurons overlaid with the average time course of mouth movements as detected with a frame-by-frame image analysis (gray traces). Shading around traces, SEM; x-axis, time (0, cue onset); y-axis for popPSTHs, auROC; y-axis for mouth movements, difference in pixel intensity for adjacent pixels. B, Scatter plot comparing the latency of cue responses for individual neurons (x-axis) with the latency of the earliest mouth movement detected in each session (y-axis). Left, Suc_cue; right, Q_cue neurons. C, Representative example of a Suc_cue neuron whose response precedes mouth movements. From top to bottom: time course of mouth movements averaged for all the Suc_cue trials; raster plots and PSTHs for a single Suc_cue-selective neuron in response to Suc_cue (left) and Q_cue (right). x-axis, Time (0, cue onset); y-axis for mouth movements, difference of intensity for adjacent pixels; y-axis for rasters, trials; y-axis for PSTHs, firing frequency. Vertical line at time 0 indicates the onset of the cue. Cyan triangle superimposed to raster plots, Time of lever press; red triangle, time of earliest mouth movement detected. Shaded vertical rectangles on PSTHs show the time window in which cue responsiveness was considered.

Figure 4.

Suc_cue responses are not related to the motor act of lever pressing. A, Population PSTH for Suc_cue neurons in response to self-administrations (cyan traces) and to erroneous lever presses (gray traces). Erroneous lever presses are defined as those that occur in the absence of the cue and that do not lead to a Suc self-delivery. Shading around traces, SEM; x-axis, time (0, time at which lever is pressed); y-axis: auROC; thick line above traces, time interval in which popPSTH for self-delivery trials is significantly different from popPSTH for erroneous trials (excitatory and inhibitory responses combined). B, Response to cued lever presses and erroneous lever presses in a representative Suc_cue-selective neuron. Top, Raster plot; bottom, PSTH; x-axis, time (vertical line at time 0, time at which lever is pressed); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; cyan marker superimposed on raster plots, onset of the cue.

Figure 5.

Cue-specific responses develop with learning. A, Cue responses in first go/no-go training sessions for rats that do not learn in the first session (i.e., no_learn sessions). Left, Representative behavioral record of a first go/no-go training session; x-axis, trials; y-axis, percentage of cues followed by a lever press; cyan line, Suc trials; gold line, Q trials; black line, performance averaged across cues; dashed line, criterion level. Middle, Distribution of cue-responsive neurons recorded in no_learn sessions. Gray dashed bars, Distribution of cue responses in well trained rats [Suc_cue-selective (cyan), Q_cue-selective (gold), differentially responding to both cues (magenta), equally responding to both cues (black)]. Dark shades, Excitatory responses; light shades, Inhibitory responses. Right, Absolute difference between PSTHs in response to Suc_cue and Q_cue averaged across neurons. x-axis, time (0, cue onset); y-axis, ΔauROC; solid line, no_learn; dashed line; well trained animals; shading around traces, SEM; thick line above traces, time interval in which popPSTH for no_learn sessions is significantly different from popPSTH for well trained sessions. B, Cue responses in first go/no-go training sessions in which rats show signs of learning (i.e., first_learn sessions). Left, Representative behavioral record of a first session go/no-go training session. Middle, Distribution of cue-responsive neurons recorded in first_learn sessions. Right, Absolute difference between PSTHs in response to Suc_cue and Q_cue averaged across neurons. Conventions are as in A. C, Within-session analysis of learning. Left, Correct performance for each quarter of no_learn (light gray) and first_learn (dark gray) sessions. Middle, Percentage of cue-responsive neurons that respond selectively to cues for each quarter of no_learn (light gray) and first_learn (dark gray) sessions. Right, Representative neuron developing cue selectivity within a single first_learn session. Top, Raster plots divided by quarters; bottom, PSTHs for each quarter; blue lines, Suc trials; gold lines, Q trials; x-axis, time (vertical line at time 0, cue onset); y-axis for rasters, trials; y-axis for PSTHs, firing frequency. Asterisks indicate significant differences.

Figure 6.

Magnitude of cue responses is significantly reduced by extinction. A, Representative behavioral record of a partial extinction session. Cyan line, Suc trials; gold line, Q trials; black line, correct performance averaged across cues; light gray shading, time period in which behavior was above criterion performance; arrow, beginning of partial extinction; dark gray shading, time period with partially extinguished performance; top dashed line, criterion level; bottom dashed line, partial extinction level (<40% of pressing to Suc_cue). B, Magnitude of Suc_cue responses before (i.e., at criterion performance) and after partial extinction. Asterisk indicates significant differences. C, Scatter plot showing cue responses before and after partial extinction for each cue-responsive neuron. A total of 78.4% of the neurons (29 of 37 neurons) shows smaller responses after extinction. Dark blue dots, Nonselective neurons; light blue dots, Suc_cue-selective neurons. D, Population PSTHs for excitatory (black lines) and inhibitory (white line) responses before (light gray shading) and after (dark gray shading) partial extinction. Pre E, Excitatory responses before extinction; Pre I, inhibitory responses before extinction; Post E, excitatory responses after extinction; Post I, inhibitory responses after extinction; x-axis, time (0, cue onset); y-axis, auROC; shading around traces, SEM. E, Representative raster plots and PSTHs in response to Suc_cue (left) and Q_cue (right) before (top) and after (bottom) extinction. x-axis, Time (vertical line at time 0, cue onset); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; cyan marker, time of self-delivery. F, Reduction of cue responses as result of time passing in a regular go/no-go session (left bar) and in length-matched extinction sessions. Error bars represent the SEM. Asterisk indicates significant differences.

Figure 7.

Suc_cue neurons have larger responses to sucrose compared with Q_cue-selective neurons. A, Population PSTH of excited Suc_cue-selective (cyan) and Q_cue-selective (gold) neurons in response to Suc self-administration. x-axis, Time (0, Suc delivery); y-axis, auROC; shading around traces, SEM; horizontal bar above traces, time interval in which responses are significantly different. B, Cumulative distribution function of responses to Suc for Suc_cue-selective (shades of blue) and Q_cue-selective (shades of gold) neurons grouped according to the strength of the cue response. x-Axis, auROC; y-axis, probability. C, Responses to Suc averaged between 250 and 500 ms and across different groups of Suc_cue-selective (shades of blue) and Q_cue-selective (shades of gold) neurons. x-axis, Groups of cue-responsive neurons according to strength of response; y-axis, auROC; error bars, SEM. D, Representative examples of a neuron selective to the Suc_cue (top row) and one selective for the Q_cue (bottom row). Raster plots (top) and PSTHs show responses to the Suc_cue (left), Q_cue (middle), and Suc taste (right). x-Axis, Time (vertical line at time 0 in Suc_cue and Q_cue rasters and PSTHs, onset of cue; vertical line at time 0 in Suc taste rasters and PSTHs, time of Suc self-delivery); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; cyan triangle, time of Suc self-delivery; gold triangle, time of Q self-delivery; cyan circle, time of cue onset.

Figure 8.

Responses to sucrose match responses to Suc_cue. A, Population PSTH in response to Suc for the following three groups of neurons: neurons with selective excitatory responses to Suc_cue (left, cyan); neurons with selective inhibitory responses to Suc_cue (middle, light blue); and neurons with no responses to either cue (right, gray). x-axis, Time (vertical dashed line at 0, Suc delivery); y-axis, auROC; shading around traces, SEM; horizontal dashed line, baseline auROC. B, Suc responses averaged across neurons and over 250 ms following taste delivery in the three groups of neurons outlined in A. Same color conventions as A. Error bars represent the SEM. Asterisks indicate significant differences. C, Left, Frequency histogram of responses to Suc for three groups of neurons. x-axis, auROC; y-axis, count; dark colors (i.e., dark blue, dark cyan, and dark gray), neurons with excitatory responses to Suc; light colors (i.e., light blue, light cyan, and light gray), neurons with inhibitory responses to Suc; white, neurons with no responses to Suc. Right, Top, Percentage of neurons with excitatory, inhibitory, and nonsignificant responses to Suc for each of the three groups described in A. x-axis, Subgroups; y-axis, percentage of total neurons responding to Suc. Right, Bottom, Cumulative distribution function for neurons with selective excitatory responses to Suc_cue (cyan), neurons with selective inhibitory responses to Suc_cue (light blue), neurons with no responses to either cue (gray). x-axis, auROC; y-axis, probability. Asterisks indicate significant differences. D, Representative raster plots and PSTHS for two matching neurons and a non-cue-responsive neuron. Top row, Neuron excited by Suc_cue and Suc; middle row, neuron inhibited by Suc-cue and Suc; bottom row, non-cue-responsive neuron; x-axis, time (vertical line at time 0 in Suc_cue and Q_cue rasters and PSTHs, onset of cue; vertical line at time 0 in Suc taste rasters and PSTHs, onset of Suc self-delivery); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; cyan triangle, time of Suc self-delivery; gold triangle, time of Q self-delivery; cyan circle, time of Suc_cue onset.

Figure 9.

Matching neurons do not code prediction error. A, Pop PSTHs for matching neurons in responses to self-delivered Suc (cyan shading; Exp E, excitatory responses; Exp I, inhibitory responses) and unexpectedly delivered Suc (gray shading; Un E, excitatory responses; Un I, inhibitory responses). x-axis, Time (dashed line at 0, Suc delivery); y-axis, auROC. Population PSTHs are plotted for both excitatory and inhibitory responses. Shading, SEM. B, Scatter plot showing responses to unexpected vs expected Suc. x-axis, auROC; y-axis, auROC. C, Representative raster plot and PSTHs for a Suc_cue-selective neuron showing responses to unexpected and expected Suc. x-axis, Time (vertical line at time 0 in left rasters and PSTHs, onset of cue; vertical line at time 0 in right rasters and PSTHs, onset of Suc delivery); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; cyan triangle, time of Suc self-delivery; cyan circle, time of Suc_cue onset.

Figure 10.

Matching neurons encode sucrose even in its absence. A, Population PSTH in response to Suc self-administrations and omissions for matching neurons (left) and non-cue-responsive neurons (right). Cyan traces in left panel, Responses to self-administered Suc in matching neurons; light gray traces in left panel, responses to omissions in matching neurons; dark gray traces in right panel, responses to self-administered Suc in non-cue-responsive neurons; light gray traces in left panel, responses to omissions in non-cue-responsive neurons; Exp E, excitatory responses to expected Suc; Exp I, inhibitory responses to expected Suc; Om E, excitatory responses to omissions; Om I, inhibitory responses to omissions; shading around traces, SEM; x-axis, time (dashed line at 0, lever press; i.e., Suc self-administration or omission); y-axis, auROC. B, Scatter plot comparing responses to Suc and omission for matching neurons. Dark cyan, Neurons with excitatory responses; light cyan, neurons with inhibitory responses; x-axis, auROC; y-axis, auROC. C, Comparison of mouth movements evoked by Suc (dark gray) and by omissions (light gray). Shading around traces, SEM; x-axis, time (0, timing of lever press); y-axis for mouth movements, difference of intensity for adjacent pixels. D, Representative examples of a matching neuron (left panels) and a non-cue-responsive neuron (right panels). Raster plots and PSTHs in response to cues, self-administered Suc, and omission. Red dashed traces superimposed on PSTHs represent the time course of mouth movements. x-axis, time (vertical line at time 0 for cue responses, cue onset; vertical line at time 0 for Suc and omissions, lever pressing time); y-axis for rasters, trials; y-axis for PSTHs, firing frequency; y-axis for mouth movements, difference of intensity for adjacent pixels; cyan triangle, time of Suc self-delivery; gold triangle, time of Q self-delivery; cyan circle, time of Suc_cue onset.