Contrast-dependent orientation discrimination in the mouse

Abstract

As an important animal model to study the relationship between behaviour and neural activity, the mouse is able to perform a variety of visual tasks, such as orientation discrimination and contrast detection. However, it is not clear how stimulus contrast influences the performance of orientation discrimination in mice. In this study, we used two task designs, two-alternative forced choice (2AFC) and go/no-go, to examine the performance of mice to discriminate two orthogonal orientations at different contrasts. We found that the performance tended to increase with contrast, and the performance at high contrast was better when the stimulus set contained a single contrast than multiple contrasts. Physiological experiments in V1 showed that neural discriminability of two orthogonal orientations increased with contrast. Furthermore, orientation discriminability of V1 neurons at high contrast was higher in the single than in the multiple contrast condition, largely due to smaller response variance in the single contrast condition. Thus, the performance of mice to discriminate orientations at high contrast is adapted to the contrast range in the stimuli, partly attributed to the contrast-range dependent capacity of V1 neurons to discriminate orientations.

Figure 1. Dependence of orientation discrimination performance on contrast in the 2AFC task.

(a) Schematic illustration of the behavioural apparatus. (b) Performance in the multiple contrast condition for an example mouse. Each color represents one session. (c) Correct rate at each contrast in the multiple contrast condition. Spearman’s rank correlation coefficient r = 1, one-tailed P = 0.008; P = 2.4 × 10⁻⁷ for ANOVA. Error bars, s.d., n = 8 mice. (d) Comparison of performance at 100% contrast between the single contrast condition and the multiple contrast condition. Dashed line, the diagonal line. Red (black) circles represent the earlier (later) single contrast condition tested before (after) the multiple contrast condition. P = 0.008 for red data points, P = 0.004 for black data points, one-tail, Wilcoxon signed rank test.

Figure 2. Dependence of orientation discrimination performance on contrast in the go/no-go task.

(a) Left, schematic illustration of the behavioural apparatus. Right, The timing of trial events and behavioural outcomes. Gray bar from 0 to 3 s indicates the duration of visual stimuli, black bar indicates the response window, and blue bar indicates the inter-stimulus interval. FA, false alarm. CR, correct rejection. (b) Discriminability in the multiple contrast condition in which the contrasts ranged from 15% to 100% (P = 0.15, ANOVA). (c) Discriminability in the multiple contrast condition in which the contrasts ranged from 15% to 80% (P < 0.05, ANOVA). Error bars, s.d., n = 14 mice. (d) Comparison of discriminability between 62% and 100% contrasts for the multiple contrast condition in which the contrasts ranged from 15% to 100%. Dashed line, the diagonal line. P = 0.001, one-tail, Wilcoxon signed rank test. (e) Comparison of discriminability between 53% and 80% contrasts for the multiple contrast condition in which the contrasts ranged from 15% to 80%. P = 4.3 × 10⁻⁴, one-tail, Wilcoxon signed rank test. (f) Comparison of discriminability at 100% contrast between the single contrast condition and the multiple contrast condition. Red (black) circles represent the earlier (later) single contrast condition tested before (after) the multiple contrast condition. P = 6.1 × 10⁻⁴ for red data points, P = 6.1 × 10⁻⁵ for black data points, one-tail, Wilcoxon signed rank test. (g) Comparison of discriminability at 80% contrast between the single contrast condition and the multiple contrast condition. The single contrast condition was measured after the multiple contrast condition. P = 6.1 × 10⁻⁵, one-tail, Wilcoxon signed rank test.

Figure 3. Hit rate, FA rate, and response bias in the multiple contrast condition during the go/no-go task.

(a,b) Hit rate increased with contrast. Spearman’s rank correlation coefficient r = 1, P = 0.008. Upper panel, the contrasts ranged from 15% to 100%; lower panel, the contrasts ranged from 15% to 80%. (c,d) FA rate increased with stimulus contrast (P < 0.05, ANOVA). (e,f) Response bias increased with stimulus contrast. Spearman’s rank correlation coefficient r = 1, P = 0.008. Error bars, s.d., n = 14 mice. ***P < 0.001, Wilcoxon signed rank test.

Figure 4. Responses of example V1 neurons to oriented gratings at different contrasts.

(a) Spike rasters of an example neuron in response to drifting gratings at different orientations and contrasts. (b) Responses to preferred and orthogonal orientations for the cell in (a). Red, contrast response function for preferred orientation. Blue, contrast response function for orthogonal orientation. Magenta, response difference between preferred and orthogonal orientations as a function of contrast. (c) Same as described in (b) for another example cell. Error bars, s.e.m.

Figure 5. Dependence of P-O response difference on contrast for V1 neurons.

(a) Upper: distribution of monotonicity index for the responses measured with contrasts ranging from 15% to 100% (n = 237, including 188 neurons from anaesthetized mice and 49 neurons from awake mice). Lower: distribution of monotonicity index for the responses measured with contrasts ranging from 15% to 80% (n = 148 neurons from anaesthetized mice). (b) Response difference between preferred and orthogonal orientations averaged over all neurons. Upper: contrast range of 15% to 100%, Spearman’s rank correlation coefficient r = 1, P = 0.008, n = 237. Lower: contrast range of 15% to 80%, Spearman’s rank correlation coefficient r = 1, P = 0.008, n = 148. Error bars, s.e.m.

Figure 6. ROC analysis and neurometric function.

(a) Distribution of responses from an example neuron in response to preferred (red bars) and orthogonal (blue bars) orientations at five contrast levels. (b) Receiver operating characteristic (ROC) curves for the responses to the five contrasts shown in (a). For each ROC curve, the probability that the response to the preferred orientation exceeds a given criterion value is plotted against that to the orthogonal orientation. (c) Neurometric function for the responses shown in (a). The ROC area for each contrast was computed from the corresponding ROC curve. (d–f), same as described in (a–c) for another example neuron.

Figure 7. Analysis of neurometric functions of V1 neurons.

(a) Distributions of the contrast at which the ROC area was maximum. Upper: the responses were measured using contrasts ranging from 15% to 100%, n = 237. Lower: the responses were measured with contrasts ranging from 15% to 80%, n = 148. (b) ROC area versus contrast, averaged over all neurons. Upper: the responses were measured using contrasts ranging from 15% to 100%, Spearman’s rank correlation coefficient r = 1, P = 0.008, n = 237. Lower: the responses were measured with contrasts ranging from 15% to 80%, Spearman’s rank correlation coefficient r = 1, P = 0.008, n = 148. Error bars, s.e.m.

Figure 8. Comparison of ROC area and neural responses to high contrast stimuli between the single and the multiple contrast conditions.

(a) ROC area at 100% contrast was significantly higher in the single than in the multiple contrast condition in anaesthetized mice. P = 0.01, n = 87, Wilcoxon signed rank test. (b) Response difference between preferred and orthogonal orientations at 100% contrast for the single and the multiple contrast conditions in anaesthetized mice. Spike count in each trial was computed as the number of spikes during the 2 s of the drifting grating. P = 0.15, n = 87, Wilcoxon signed rank test. (c,d) Response variance to preferred (P) or orthogonal (O) orientation at 100% contrast was significantly smaller in the single than in the multiple contrast condition in anaesthetized mice. P = 0.002 and 4.4 × 10⁻⁷, respectively, n = 87, Wilcoxon signed rank test. (e–h) Results for responses measured at 80% contrast in anaesthetized mice, similar as those described in (a–d). n = 92. (i–l) Results for responses measured at 100% contrast in awake mice, similar as those described in (a–d). n = 17. Error bars, s.e.m., *P < 0.05, **P < 0.01, ***P < 0.001, Wilcoxon signed rank test.