Task specificity in mouse parietal cortex

Abstract

Parietal cortex is implicated in a variety of behavioral processes, but it is unknown whether and how its individual neurons participate in multiple tasks. We trained head-fixed mice to perform two visual decision tasks involving a steering wheel or a virtual T-maze and recorded from the same parietal neurons during these two tasks. Neurons that were active during the T-maze task were typically inactive during the steering-wheel task and vice versa. Recording from the same neurons in the same apparatus without task stimuli yielded the same specificity as in the task, suggesting that task specificity depends on physical context. To confirm this, we trained some mice in a third task combining the steering wheel context with the visual environment of the T-maze. This hybrid task engaged the same neurons as those engaged in the steering-wheel task. Thus, participation by neurons in mouse parietal cortex is task specific, and this specificity is determined by physical context.

Keywords: cerebral cortex; decision-making; sensorimotor processing.

Figure 1

In mice performing two visual decision tasks, many parietal neurons are task specific (A) Top: the T-maze task. Bottom: fraction of rightward choices versus contrast of stimuli on the left (negative) or on the right (positive). Dots and error bars show mean ± SD for n = 21 sessions in 6 mice. Curves show the fitted psychometric function for each mouse (gray) and averaged across mice (orange). Psychometric data for all sessions are in Figure S1 (B) Top: the steering-wheel task. Bottom: performance in the task of the same mice on the same days as (A). (C) Top: map of visual cortical areas from wide-field imaging, showing the visual field sign of retinotopic areas (blue: negative; red: positive) and the field of view targeted for two-photon imaging (inset) from an example mouse. Bottom: outlines of the identified neurons in the field of view. (D) Responses of two neurons from the example session, showing task-specific activity. (E) Summary of activity (isolation distance) in the example session in the T-maze (TM) versus steering-wheel (SW) tasks, showing neurons that fired only in the T-maze task (orange), only in the steering-wheel task (blue), in both tasks (white), or in neither task (gray). Diamonds indicate the example neurons in (D). (F) Raster plots of neurons in an example session in the two tasks. Gray level denotes deconvolved calcium signal, Z scored. Neurons are sorted by relative task preference, i.e., the difference between the x and y values in (E). (G) Anatomical distribution for the same example mouse, showing the overlay of ROIs over nine sessions, colored as in (E). Dashed square indicates a typical imaging field of view as in (C). Scale bar indicates extent of task selectivity (difference of task activity as in [E] normalized over the sum), averaged over sessions.

Figure 2

Task specificity is consistent across days (A) Activity of two example neurons in the T-maze on consecutive days (left). Activity of the same neurons in the steering-wheel task across days (right). Insets show the mean images of each neuron in each session. Each bar represents 10 min. (B) Comparison of activity within tasks across consecutive days, in the T-maze (left) or steering-wheel task (right). Correlations were positive in both cases (r = 0.83 and r = 0.77, p ≈ 0, i.e., too small to measure). (C) Same as in (B) but comparing activity across tasks. Correlations were negative (left: r = −0.24, p = 0.08) or not significant (right: r = −0.27, p = 0.05). (D) Comparison of task preference (relative activity over tasks: positive for neurons preferring the T-maze task and negative for neurons preferring the steering-wheel task) for neurons imaged in two example consecutive days (N = 56 cells), showing significant correlation across days, r = 0.84, p = 5e−16. Correlations were also high in the other three pairs of days, with r = 0.85, 0.87, and 0.78. (E) Summary from four pairs of days in three mice. Diamond illustrates the example pair of days from (B) and (C). Filled points indicate significant Spearman rank correlations at p < 0.05. (F) Spearman rank correlation across all conditions, for the example pair of days in (B) and (C) (left) and the average over four pairs of days from (E) (right). Scale bar indicates Spearman rank correlation, rho (ρ).

Figure 3

Task specificity is predictable by physical context in the absence of a task (A) Raster plot of activity from neurons in an example session showing 5-min segments of activity in each task and in the corresponding passive condition. Gray level indicates normalized firing rate as in Figure 1. Left to right: T-maze, passive ball, steering-wheel task, and passive steering wheel. (B) Comparison of activity for the same population of neurons across conditions with similar physical context, for the example session in Figures 2A–2C. Activity is highly correlated both within the ball context (left: r = 0.63, p ≈ 0) and within the wheel context (right: r = 0.65, p ≈ 0). (C) Comparison of activity across different physical contexts for the same session. Activity is not significantly correlated (left: r = −0.16, p = 0.09; right: r = −0.10, p = 0.28). (D) Summary of correlations of activity within and across physical contexts for 10 sessions where we recorded passive conditions. Filled circles indicate significant Spearman rank correlations. Correlations were different across but not within contexts, one-way ANOVA, F(3,36) = 9.43, p = 1e−16. (E) Another pair of sessions where all four conditions were recorded on successive days. Color map and scale bar shows Spearman rank correlation, rho (ρ), of activity as before.

Figure 4

Activity in a hybrid task confirms the role of physical context (A) The “steering T-maze” task (STM) combines the apparatus of the steering wheel with the visual scene of the T-maze in a fixed position along the corridor. (B) Three example neurons from a session that included all three tasks. (C) Activity of the same population of neurons across the steering-wheel and steering T-maze tasks (left: r = 0.74, p ≈ 0) and across the T-maze and steering T-maze tasks (right: r = −0.02, p = 0.72) from the same session as (B). (D) Summary of pairwise comparisons between the T-maze, steering-wheel, and hybrid tasks. SW versus hybrid: n = 9 sessions; SW versus TM: n = 3 sessions; TM versus hybrid: n = 3 sessions. Sessions were only included if mice performed all three tasks, and enough trials were acquired for each pair. A one-way ANOVA found significant group differences, F(2,12) = 17.74, p = 0.0003.