Beneficial effects of the NMDA antagonist ketamine on decision processes in visual search

Abstract

The ability of sensory-motor circuits to integrate sensory evidence over time is thought to underlie the process of decision-making in perceptual discrimination. Recent work has suggested that the NMDA receptor contributes to mediating neural activity integration. To test this hypothesis, we trained three female rhesus monkeys (Macaca mulatta) to perform a visual search task, in which they had to make a saccadic eye movement to the location of a target stimulus presented among distracter stimuli of lower luminance. We manipulated NMDA-receptor function by administering an intramuscular injection of the noncompetitive NMDA antagonist ketamine and assessed visual search performance before and after manipulation. Ketamine was found to lengthen response latency in a dose-dependent fashion. Surprisingly, it was also observed that response accuracy was significantly improved when lower doses were administered. These findings suggest that NMDA receptors play a crucial role in the process of decision-making in perceptual discrimination. They also further support the idea that multiple neural representations compete with one another through mutual inhibition, which may explain the speed-accuracy trade-off rule that shapes discrimination behavior: lengthening integration time helps resolve small differences between choice alternatives, thereby improving accuracy.

Figure 1.

Visual search task and behavior. A, Visual search task trial progression. The target is defined as the stimulus with higher contrast (contrast increment task). The arrow represents a correct response in which a single saccade is made to the target. B, Eye position traces from a single session (monkey G) during trials in which a saccade was directed to the search target before (thin lines) and after (thick lines) a 0.5 mg/kg ketamine injection.

Figure 2.

Representative effects of ketamine on visual search performance (monkey G, 0.5 mg/kg ketamine). A, Mean (±SE) saccade amplitude for 1 min intervals following the injection of ketamine. The gray shaded area indicates the epoch during which there was a significant decrease in saccade amplitude. The mean (including ±SE) saccade amplitude for the control block is indicated by the horizontal line. B, Mean (±SE) response latency for each minute following the injection of ketamine. C, Mean (±SE) response latency as a function of stimulus luminance difference before and after ketamine administration. D, Mean (±95% CI) response accuracy as a function of stimulus luminance difference before and after ketamine administration. Ketamine increased the proportion of correct trials at intermediate stimulus luminance differences (see Materials and Methods) (p < 0.05, χ² test), shifting the psychometric function to the left. Discrimination threshold was taken as the point at which the Weibull function reached 64% of its maximum (dashed line).

Figure 3.

Changes in response latency and discrimination threshold following ketamine and control treatments. A, Percentage (±95% CI) change in response latency for each animal and for each treatment, compared with corresponding control data (monkey G, 151 ± 0.4 ms; monkey F, 212 ± 0.8 ms; monkey H, 178 ± 0.7 ms). All percentage changes were significantly different from 0 (p < 0.05, t test). B, Percentage change in discrimination threshold for each animal and for each treatment, compared with corresponding control data (monkey G, 8.0 cd/m² or 43% contrast difference; monkey F, 6.0 cd/m² or 36.1%; monkey H, 9.2 cd/m² or 46.5%). Bars with black outlines indicate a significant difference in accuracy in at least one of the intermediate stimulus luminance differences (see Materials and Methods) (p < 0.05, χ²). Insets illustrate the shift in the psychometric function (axis labels as in Fig. 2D) for three example sessions (monkeys G and F, 0.25 mg/kg; monkey H, 0.5 mg/kg). Treatments with no injection (no inj) and saline injections (sal) are identified as treatments with 0 mg/kg ketamine dose.

Figure 4.

Difference between the response latency in treatment trials and that predicted by the sum of the mean response latency in the corresponding control session and at the highest contrast difference (86.1%). Data from each animal is indicated by a different symbol (triangle, monkey G; square, monkey F; circle, monkey H). Response latency differences were determined to be statistically significant (solid symbols) if they exceeded by 2 SD the mean latency difference calculated in the saline sessions (black symbols and gray areas). Inset, Schematic of the calculation of response latency difference (shaded area): RLD_x = TL_x − [CL_x + (TL_86.1% − CL_86.1%)], where RLD_x is the response latency difference at X luminance difference; TL_x and CL_x are the mean response latencies at X luminance difference in the treatment and control blocks, respectively; and TL_86.1% and CL_86.1% are the mean treatment and control block latencies at the highest luminance difference. Inset example is from Figure 2C.

Figure 5.

Response accuracy plotted as a function of mean response latency before (○) and after (●) treatment with ketamine. Data are from the trials with intermediate levels of stimulus luminance difference (20.9, 36.5, and 48.0% contrast differences) for which the ketamine dose led to a significantly lower discrimination threshold (Fig. 3B). Black lines indicate significant increases in response accuracy (p < 0.05, χ² tests). All data shown had significant increases in response latency (p < 0.05, rank sum tests). Average change in accuracy and latency was 0.11 (range, 0.004–0.24) and 34.7 ms (range, 6–82), respectively.