Distinct Waking States for Strong Evoked Responses in Primary Visual Cortex and Optimal Visual Detection Performance

Abstract

Variability in cortical neuronal responses to sensory stimuli and in perceptual decision making performance is substantial. Moment-to-moment fluctuations in waking state or arousal can account for much of this variability. Yet, this variability is rarely characterized across the full spectrum of waking states, leaving the characteristics of the optimal state for sensory processing unresolved. Using pupillometry in concert with extracellular multiunit and intracellular whole-cell recordings, we found that the magnitude and reliability of visually evoked responses in primary visual cortex (V1) of awake, passively behaving male mice increase as a function of arousal and are largest during sustained locomotion periods. During these high-arousal, sustained locomotion periods, cortical neuronal membrane potential was at its most depolarized and least variable. Contrastingly, behavioral performance of mice on two distinct visual detection tasks was generally best at a range of intermediate arousal levels, but worst during high arousal with locomotion. These results suggest that large, reliable responses to visual stimuli in V1 occur at a distinct arousal level from that associated with optimal visual detection performance. Our results clarify the relation between neuronal responsiveness and the continuum of waking states, and suggest new complexities in the relation between primary sensory cortical activity and behavior.SIGNIFICANCE STATEMENT Cortical sensory processing strongly depends on arousal. In the mouse visual system, locomotion (associated with high arousal) has previously been shown to enhance the sensory responses of neurons in primary visual cortex (V1). Yet, arousal fluctuates on a moment-to-moment basis, even during quiescent periods. The characteristics of V1 sensory processing across the continuum of arousal are unclear. Furthermore, the arousal level corresponding to optimal visual detection performance is unknown. We show that the magnitude and reliability of sensory-evoked V1 responses are monotonic increasing functions of arousal, and largest during locomotion. Visual detection behavior, however, is suboptimal during high arousal with locomotion, and usually best during intermediate arousal. Our study provides a more complete picture of the dependence of V1 sensory processing on arousal.

Keywords: arousal; locomotion; pupil; state; visual cortex; visual detection.

Figure 1.

Waking state, as indexed by pupil size, varies over a wide range. A, Basic experimental setup for the head-fixed mouse preparation. An LCD monitor is mounted to the mouse's left, presenting visual stimuli to the left eye, and a USB camera is mounted to the mouse's right, capturing video of the right eye, which is illuminated by 2 IR LEDs. A photodiode mounted to the upper-left corner of the LCD monitor gives precise timestamps of stimulus onsets. An optical mouse mounted to the side of the wheel acquires wheel speed. B, Representative traces of normalized pupil diameter and locomotion speed for an entire example recording session (∼90 min). The horizontal dashed line plotted with wheel speed indicates the 5 cm/s threshold for qualifying as a locomotion bout (see Materials and Methods, “State-dependent electrophysiological metrics”). Insets magnify the numbered regions and show regions of interest (ROIs) around the eye for selected times. C, Normalized pupil diameter distributions, sorted according to whether mice were still or walking, for all task-engaged (N = 20) or passively behaving (N = 25) animals in the dataset. Vertical dotted lines indicate the peaks of the two distributions.

Figure 2.

Spontaneous firing rates in V1 are minimal during intermediate arousal without locomotion. A, Example traces of MUA simultaneously recorded in layer 2/3 and 5 of V1, along with layer 5 LFP, normalized pupil diameter, and locomotion speed. B, Spontaneous firing rates (as a fraction of the maximum spontaneous firing rate recorded during a session) in layer 5 (left) and 2/3 (right) as a function of baseline pupil diameter and sorted by locomotion status. C, Histograms of layer 5 (left) and layer 2/3 (right) extracellular MUA recordings, sorted into bins of pupil diameter in which the minimum spontaneous firing rates occurred during an individual recording. 95% Confidence intervals (CIs) are from bootstrap resampling (10,000 times) of values within bins. Baseline pupil diameter bin widths were chosen such that an equal amount of data fell into each bin. (layer 5: n = 17 animals, n = 23 recordings; layer 2/3: n = 6 animals, n = 6 recordings; data are shown as mean ± 68% CI).

Figure 3.

Visually evoked spiking responses in V1 are enhanced monotonically with increasing arousal, and largest during locomotion. A, Example PSTH of the evoked layer 5 MUA response to a full-contrast Gaussian noise movie. B, Evoked layer 5 multiunit firing rate (as a fraction of spontaneous, baseline firing rate 500 ms before stimulus presentation) in response to full-contrast Gaussian noise movies, as a function of baseline pupil diameter, and sorted by locomotion status. C, Trial-by-trial reliability (cross-correlation, c.c.) of multiunit spiking responses to full-contrast Gaussian noise movies as a function of baseline pupil diameter, and sorted by locomotion status. “Raw” denotes the pairwise cross-correlation between evoked spiking responses to the same Gaussian noise movie, and “chance” denotes the pairwise cross-correlation between evoked spiking responses and periods of spontaneous spiking activity occurring in the same pupil diameter bin, a correction for cross-correlation increases due to increased spiking alone. D, Within-recording comparisons of evoked layer 5 multiunit firing rate (fraction of baseline firing rate) between still and locomotion periods during high arousal (i.e., in pupil diameter bins in which locomotion occurred). p-value from rank-sum test. E, Within-recording comparisons of evoked layer 5 multiunit spike reliability (trial-by-trial cross-correlations of evoked PSTHs) between still and locomotion periods during high arousal. p-value from rank-sum test. F, Top, Histogram of extracellular MUA recordings, sorted into bins of pupil diameter (during stillness) in which the largest evoked responses occurred during an individual recording. Bottom, Same histogram, but including locomotion periods. G, As in F, but for highest evoked spike reliability. For F and G, 95% CIs are from bootstrap resampling (10,000 times) of values within bins. For pupil diameter bins in B, C, F, and G, bin widths were chosen such that an equal amount of data fell into each bin (N = 17 animals, n = 23 recordings; data are shown as mean ± 68% CI).

Figure 4.

State dependence of evoked responses in V1 layer 2/3 is similar to that of layer 5. A, Representative CSD plot used to localize silicon probe contacts residing in layer 2/3. Within ∼40 ms of a 50 ms full-screen flash, a strong, short-latency sink is evident in mid-layers (arrow), followed by delayed sinks in more superficial and deep layers after 50 ms. Even stronger sinks are evident in putative deep layer 5 100 ms after stimulus onset, likely due to polysynaptic activity induced by the stimulus. Contacts in layer 2/3 were considered to be 50–100 μm above the estimated layer 4 boundary. B, Evoked layer 2/3 multiunit firing rate (as a fraction of spontaneous, baseline firing rate 500 ms before stimulus presentation) in response to full-contrast Gaussian noise movies, as a function of baseline pupil diameter, and sorted by locomotion status. C, Trial-by-trial reliability (cross-correlation, c.c.) of layer 2/3 multiunit spiking responses to full-contrast Gaussian noise movies as a function of baseline pupil diameter, and sorted by locomotion status. “Raw” denotes the pairwise cross-correlation between evoked spiking responses to the same Gaussian noise movie, and “chance” denotes the pairwise cross-correlation between evoked spiking responses and periods of spontaneous spiking activity occurring in the same pupil diameter bin, to correct for cross-correlation increases due to increased spiking alone. For pupil diameter bins in B and C, bin widths were chosen such that an equal amount of data fell into each bin (N = 6 animals, n = 6 recordings; data are shown as mean ± 68% CI).

Figure 5.

Complete artificial dilation of the visually stimulated eye does not fundamentally alter the enhancement of evoked V1 responses as a function of pupil size. A, Individual video frames from sessions in which two cameras were positioned to image each eye. Before atropine application, the pupil sizes of both eyes fluctuate coherently. After atropine application to the left eye, the pupil of the right eye still varies with state, but the pupil of the left eye is always fully dilated. B, Probability histogram of normalized pupil diameter for animals treated with atropine in the visually stimulated eye (top), compared with control animals under the same passive-behaving conditions (bottom, distribution same as in Fig. 1C). While normalized pupil diameter spans the full range in atropine-treated animals, the distribution is notably centered on smaller values. Vertical dotted lines indicate the peaks of the two distributions. C, Evoked firing rate and spike reliability as a function of baseline pupil diameter and sorted by locomotion status for atropine-treated animals. D, Histograms (counts = number of animals) of pupil diameter bins (still and locomotion periods combined) associated with the largest evoked firing rate recorded per animal (top) and the highest evoked spike reliability recorded per animal (bottom), compared between atropine-treated animals and control animals (control animals same as those reported in Fig. 3). p-values are from Fisher's exact test. For pupil diameter bins in C and D, bin widths were chosen such that an equal amount of data fell into each bin (N = 17 animals, N = 17 recordings; data are shown as mean ± 68% CI).

Figure 6.

V_m of V1 cortical neurons exhibits depolarization and decreased variability as a function of arousal, with most depolarized and least variable V_m during locomotion. A, Example traces of V_m from a V1 layer 5 regular-spiking, putative pyramidal cell, pupil diameter, and locomotion speed. Inset 1 emphasizes V_m changes associated with the transition from stillness to locomotion. Inset 2 emphasizes the relationship between pupillary microdilations (arrows) and V_m depolarizations during stillness. Spikes are truncated in example traces to emphasize subthreshold behavior. B, Coherence between pupil diameter and V_m. Coherence is high at low frequencies (0.01–0.02 Hz), with a smaller secondary peak from ∼0.1–0.3 Hz, likely reflecting the tracking of pupillary microdilations by V_m. C, Cross-correlation between pupil diameter and V_m, showing a lag of ∼1 s between changes in V_m and associated changes in pupil diameter. D, Mean spontaneous V_m (expressed as a change in V_m from the most hyperpolarized V_m recorded during a session) as a function of baseline pupil diameter and sorted by locomotion status. E, V_m SD as a function of baseline pupil diameter and sorted by locomotion status. F, V_m Hilbert amplitude at 2–10 Hz and 50–100 Hz as a function of baseline pupil diameter and sorted by locomotion status. G, Density plot, pooling V_m values recorded from all cells in the dataset, encapsulating data shown in D and E. Each vertical slice of the plot is the probability distribution of ΔV_m (the difference between V_m in a recorded cell and the most hyperpolarized V_m recorded in that cell) in a given pupil diameter range. A more dispersed probability distribution is evident at smaller pupil diameters, while a narrower distribution is evident at larger pupil diameters. For pupil diameter bins in D–G, bin widths were chosen such that an equal amount of data fell into each bin (N = 8 animals, n = 10 cells; data are shown as mean ± 68% CI).

Figure 7.

Performance of a target-in-noise visual detection task is suboptimal during locomotion. A, Trial structure of the task. Each trial begins with a variable-duration foreperiod consisting of a sequence of Gaussian noise movies, during which the mouse must withhold licking. Mice must lick during the target period, in which a drifting square-wave grating is embedded in one of the noise movies with different blend ratios, modulating task difficulty. Note that for display purposes, the aspect ratio of the visual stimulus screen in the figure is different from that of the actual LCD monitor. B, Overall performance rates (false alarm rates and hit rates) across animals for the different target levels presented during behavior sessions. C, Overall perceptual sensitivities (d′) across animals for the different target levels presented during behavior sessions. D, Muscimol block of V1 activity impairs detection task performance (n = 6 animals). Pre, Baseline performance before injections; sal, saline injection; musc, muscimol injection; rec, recovery from muscimol injections. After muscimol injection, but not saline injection, detection performance decreased to chance levels (d′ = 0), or slightly below chance levels. All but one animal returned to baseline performance following recovery from muscimol. p-values are from Kruskal–Wallis test. After Bonferroni's correction, significant p-values at the 0.05 level must be lower than 0.008. E, Performance rate, perceptual sensitivity (d′), and decision bias (c) as a function of baseline pupil diameter, and sorted by locomotion status. F, Within-animal comparisons of d′ for stillness versus locomotion during high arousal (i.e., pupil diameter bins in which locomotion occurred). p-value is from rank-sum test. G, Within-animal comparisons of the largest d′ prime recorded during stillness versus locomotion. Data points are colored according to the pupil diameter bin in which they were recorded. p-value is from rank-sum test. H, Histograms (counts = number of animals) of the pupil diameter bin in which the largest d′ was recorded for each animal during stillness (top) and during stillness and locomotion combined (bottom). For pupil diameter bins in E and H, bin widths were chosen such that an equal amount of data fell into each bin. For E and H, 95% CIs are from bootstrap resampling (10,000 times) of values within bins (N = 12 animals, n = 81 sessions; data are shown as mean ± 68% CI).

Figure 8.

Visual detection of Gaussian noise movies is suboptimal during locomotion. A, Trial structure of the task. Each trial begins with a variable-duration foreperiod (signaled by an audible tone) consisting of an isoluminant gray screen, during which the mouse must withhold licking. Mice must lick during the target period, in which a Gaussian noise movie is presented at different contrasts. Note that for display purposes, the aspect ratio of the visual stimulus screen in the figure is different from that of the actual LCD monitor. B, Overall performance rates (false alarm rates and hit rates) across animals for the different target levels presented during behavior sessions. C, Overall perceptual sensitivities (d′) across animals for the different target levels presented during behavior sessions. D, Performance rate, perceptual sensitivity (d′), and decision bias (c) as a function of baseline pupil diameter, and sorted by locomotion status. E, Within-animal comparisons of d′ for stillness versus locomotion during high arousal (i.e., pupil diameter bins in which locomotion occurred). p-value is from rank-sum test. F, Within-animal comparisons of the largest d′ prime recorded during stillness versus locomotion. Data points are colored according to the pupil diameter bin in which they were recorded. p-value is from rank-sum test. G, Histograms (counts = number of animals) of the pupil diameter bin in which the largest d′ was recorded for each animal during stillness (top) and during stillness and locomotion combined (bottom). For pupil diameter bins in D and G, bin widths were chosen such that an equal amount of data fell into each bin. For D and G, 95% CIs are from bootstrap resampling (10,000 times) of values within bins. (N = 8 animals, n = 59 sessions; data are shown as mean ± 68% CI).

Figure 9.

Pupil diameters associated with large and reliable visually evoked V1 responses are significantly larger than those associated with optimal visual detection performance. A, Histograms (counts = number of animals) of pupil diameter bins during stillness associated with, left to right, largest visually evoked firing rate, spike reliability, and visual detection d′ (combined from the two visual detection tasks). Largest evoked firing rate and spike reliability recorded per animal were significantly more concentrated in bins of larger pupil diameter than the largest d′ recorded per animal (p-values from Fisher's exact test; after Bonferroni correction, significant p-values at the 0.05 level must be lower than 0.017). For largest d′ recorded per animal, values were most strongly concentrated in pupil diameter bins ranging from 40 to 60% constricted from the maximal pupil diameter. B, As in A, but for still and locomotion periods combined. For pupil diameter bins, bin widths were chosen such that an equal amount of data fell into each bin.