Representation of comparison signals in cortical area MT during a delayed direction discrimination task

Abstract

Visually guided behavior often involves decisions that are based on evaluating stimuli in the context of those observed previously. Such decisions are made by monkeys comparing two consecutive stimuli, sample and test, moving in the same or opposite directions. We examined whether responses in the motion processing area MT during the comparison phase of this task (test) are modulated by the direction of the preceding stimulus (sample). This modulation, termed comparison signal, was measured by comparing responses to identical test stimuli on trials when it was preceded by sample moving in the same direction (S-trials) with trials when it was preceded by sample moving in a different direction (D-trials). The test always appeared in the neuron's receptive field (RF), whereas sample could appear in the RF or in the contralateral visual field (remote sample). With sample in-RF, we found three types of modulation carried by different sets of neurons: early suppression on S-trials and late enhancement, one on S-trials, and the other on D-trials. Under these conditions, many neurons with and without comparison effects exhibited significant, choice-related activity. Response modulation was also present following the remote sample, even though the information about its direction could only reach MT indirectly via top-down influences. However, unlike on trials with in-RF sample, these signals were dominated by response suppression, shedding light on the contribution of top-down influences to the comparison effects. These results demonstrate that during the task requiring monkeys to compare two directions of motion, MT responses during the comparison phase of this task reflect similarities and differences between the two stimuli, suggesting participation in sensory comparisons. The nature of these signals provides insights into the operation of bottom-up and top-down influences involved in this process.

Fig. 1.

Behavioral tasks, visual stimuli, and behavioral performance. A: the diagrams outline the temporal sequences of the events in a single trial. The sample and test stimuli were placed either in the neuron's receptive field (RF; top plot) or were spatially separated (bottom plot) so that the test always appeared in the RF, whereas the sample was placed at a location contralateral and noncorresponding to the RF. The monkeys fixated a spot for 1000 ms before being presented with 2 stimuli—sample and test—lasting 500 ms each and separated by a 1500-ms delay. They indicated, by pressing 1 of 2 pushbuttons, whether the sample and test moved in the same or in different directions. The rectangle around the test component of the trials highlights the portion of the trial analyzed in this paper. B: random-dot stimuli. The stimuli consisted of random dots displaced in directions chosen from a predetermined distribution. The width of this distribution determined the range of directions within which individual dots move and was varied between 0° (all dots moving in the same direction) and 360° (dots moving in all directions). C: average representative psychometric functions based on 6 randomly chosen recording sessions during sample in-RF (solid circles) and remote sample (open circles) conditions. The arrows indicate direction-range thresholds (75% correct) calculated from the best-fitting psychometric functions. Each session consisted of at least 200 trials.

Fig. 2.

Average test responses of 4 example neurons during sample moving in the same direction (S-trials) and different direction (D-trials). A and B: sample in-RF condition. A: neuron with stronger test response on D-trials (D) than on S-trials (S); D > S effect. B: neuron with stronger test response on S- than on D-trials; S > D effect. C and D: remote sample condition. C: neuron with stronger response on D- than on S-trials; D > S effect. D: neuron with stronger response on S- than on D-trials; S > D effect. Average activity is shown as a spike-density function with the Gaussian profile of σ = 20 ms. Shaded areas indicate the presence of the stimulus. Only trials with sample containing 0° range motion and test moving in the preferred direction were included. Black bars along the x-axis indicate periods where responses during the 2 sets of trials were significantly different (P < 0.05 using permutation test). Thin lines indicate ± SE. sp/s, Spikes per second.

Fig. 3.

Comparison effects during sample in-RF condition. A: differences in test responses of individual neurons during D- and S-trials computed with receiver operating characteristic (ROC) analysis (n = 171). The plot shows the strength and the time-course of these differences. Each horizontal line represents activity of a single neuron, with colors presenting area under the ROC (AROC) values calculated by comparing test activity during S- and D-trials following 0° range sample. Cells have been ordered according to the sign of the comparison effect (S > D in blue; D > S in red) and the time when the first epoch met the criterion value >0.65 (S > D) or <0.35 (D > S) cells. Triangles on the right indicate the cells with significant effects determined by bootstrap analysis (P < 0.05). The color scale bar shows the key to AROC values. B and C: distribution of maximum comparison effects for D > S and D < S cells. Note a bimodal distribution of D > S cells, early and late. D–F: average activity during S (solid lines)- and D (dotted lines)-trials of cells in the 3 groups: early D > S (n = 34), late D > S (n = 27), and S > D (n = 32). The differences between the 2 response curves, computed with ROC analysis, are shown below, separately for each group of cells. The curves show the strength and the time-course of comparisson effects in the 3 groups of cells. Thin lines indicate ± SE. G and H: effect of direction range on comparison signals during early (G) and late (H) response. Error bars ± SEM. There was no significant effect of range on the comparison signal for the 3 groups of cells, neither early in the response (50–250 ms; early D > S: P = 0.74; late D > S: P = 0.15; S > D: P = 0.62, two-sample t-test) nor late in the response (300–500 ms; early D > S: P = 0.60; late D > S: P = 0.10; S > D: P = 0.52). Data points significantly different from the AROC of 0.5 are shown by asterisks; **P < 0.01; *P < 0.05 (one-sample t-test).

Fig. 4.

Comparison effects during remote sample condition. A: differences in test responses of individual neurons during D- and S-trials computed with ROC analysis (n = 97). For details, see the legend in Fig 3A. B and C: distribution of maximum comparison effects for D > S and D < S cells. D and E: average activity during S (solid lines)- and D (dotted lines)-trials for the 2 groups of cells: D > S (n = 13) and S > D (n = 19) neurons. The differences between the 2 response curves, computed with ROC analysis, are shown below, separately for each group. The curves show the strength and the time-course of comparison effects in the 2 groups of cells. F and G: effect of direction range in the sample on comparison effects during early (F) and late (G) response. Error bars ± SEM. There was a significant effect of range for the D > S cells early in response (50–250 ms; P = 0.0004, two-sample t-test), but this was not the case for the S > D cells (P = 0.61). Late in the response, no effect of range was found for either group (300–500 ms; D > S: P = 0.58; S > D: P = 0.68). Data points significantly different from the AROC of 0.5 are shown by asterisks; **P < 0.01; *P < 0.05 (one-sample t-test).

Fig. 5.

Response modulation underlying comparison signals. Response modulation index (MI) was computed by comparing test responses with responses to the sample, MI = (R_test − R_sample)/(R_test + R_sample), with R as the firing rate. Values <0 indicate response suppression (shaded); the values >0 indicate response enhancement. A–C: sample in-RF condition. Average MIs during S- and D-trials for (A) S > D cells (n = 32), (B) late D > S cells (n = 34), and (C) early D > S cells (n = 27). D and E: remote sample condition. Average MIs during S- and D-trials for (D) S > D cells (n = 25) and (E) D > S (n = 19) cells. Error bars ± SEM; *P < 0.05; **P < 0.01 (one-sample t-test).

Fig. 6.

Correlation between response modulation during the 2 sample conditions during early (A) and late (B) responses. Each data point represents MI of an individual neuron recorded during sample in-RF and the remote sample conditions (n = 76). The data for S- and D-trials are shown by solid and open circles, respectively. The correlation for both D- and S-trials during both time epochs was significant (P < 0.05).

Fig. 7.

Comparison signals and the behavioral choice. (A–F) Sample in-RF condition. A: choice probability (CP) for cells with comparison effects plotted separately for S > D cells (blue line, n = 18), early D > S cells (red solid line, n = 26), and late D > S cells (red broken line, n = 18). B–D: distribution of CPs for S > D cells [(B) 300–500 ms epoch], early D > S cells [(C) 420–500 ms epoch], and late D > S cells [(D) 420–500 ms epoch]. E: CPs for cells with no comparison effects (n = 67). F: distribution of CPs for D = S cells during 200–400 ms epoch. B–E: thick lines along the x-axis indicate epochs with CP distributions. G–K: remote sample condition. G: average CP for cells with comparison effects plotted separately for S > D cells (blue curve, n = 19) and D > S cells (red curve, n = 13). H and I: distribution of CPs for (H) S > D and (I) D > S cells during 100–500 ms epoch. J: CP for D = S cells (n = 47). K: distribution of CPs for D = S cells during 400–500 ms epoch. Faint, thin lines indicate ± SE (A, E, G, and J). The small arrows are placed to indicate average CP (B–D, F, H, I, and K). *P < 0.05; **P < 0.01. See text for further details.