Dynamic Contextual Modulation in Superior Colliculus of Awake Mouse

Expression of spatial and temporal gain controls in neurons in SC of awake mouse. A–D, Responses of four representative neurons. The left panel in each case shows the average firing rate of the neuron during a 2-s presentation of a circular patch of drifting grating at a spatial frequency near the preferred for the neuron (0.05–0.07 cycles/^o), and centered on the receptive field. The right panel in each case shows the PSTH (bin width 0.016 s) during presentation of a patch of grating near the preferred size for that neuron, which is indicated by the arrow in the left panel. Dashed horizontal lines show the maintained rate in absence of patterned visual stimulus. Solid line shows the predictions of the size-tuning model described in Materials and Methods (Eq. 2). Error bars show ±1 SEM across trials. The insets show a spatial map of responses (white indicates no activity and darker colors indicate stronger responses) to a black square, 15° wide, flashed at each of 81 positions on the monitor. Calibration bars are 26°. The schematic below panel A shows the relative size of two patches of grating and how a grating of 0.05 cycles/^o would appear in each of them.

Correlated variability in surround suppression and adaptation effects. A, Population size-tuning for patches of drifting gratings. Each row of the image shows the predictions of the size-tuning model for a single neuron (as in Fig. 1), normalized to its maximum response. Only units in which the normalized log-likelihood of the model was at least 0.5 are shown. The units are ordered, from bottom-to-top, by the preferred size. B, Mean size-tuning for a drifting grating, obtained by averaging across the rows in A. Dashed lines show ±1 SEM across neurons. Dashed horizontal line shows the maintained rate, normalized to the unit’s maximum visual response before averaging. Arrow indicates the definition of the SI, which is the proportional reduction in response from a grating of preferred size to a large grating. C, Distribution of the SI across the population of units in A. The filled bars show SI for neurons with preferred diameter >30°. D, Population time course for drifting gratings of preferred size. The units are ordered, from bottom-to-top, by the AI. Color bar as in A. E, Mean time course for a drifting grating, obtained by averaging across the rows in D. Conventions as in B. Arrow indicates the definition of the AI, which is the proportional reduction in response from the first 0.5 s to the last 0.5 s. F, Distribution of the AI across the population of units in D. G, Comparison of the SI and AI in individual neurons (n = 93). Dashed line is the best linear fit to the data.

Surround suppression is susceptible to adaptation. A, Time course of population response for gratings for small (20° diameter) and large (90°) patches of drifting grating. Responses were normalized to the mean response across all patch sizes (most of which are not shown) before averaging (n = 98). Dashed horizontal line shows the maintained rate in absence of patterned visual stimulus, normalized in the same way. Error bars are omitted for clarity. B, Size-tuning for early (0–0.5 s) and late (1.5–2 s) response, normalized in the same way as A. Error bars are ±1 SEM across neurons that passed criteria for inclusion (early: n = 92; late: n = 80). C, Comparison of SI for early and late response (n = 73). Dashed line shows the unity line. Points falling below the line indicate neurons in which suppression was stronger in the early response than in the late response.

Tuning of surround suppression in SC of awake mouse. A, B, Response of an example neuron. A, Tuning of spiking activity evoked by a large patch of drifting grating (45° diameter) of varying orientation/direction. Dashed horizontal line shows the maintained rate in absence of patterned visual stimulus. B, Tuning of suppression induced by an annular grating of varying orientation/direction. Responses are shown for presentation of a 15° patch of drifting grating (upper dashed line, “center alone”) of direction 180°, the same stimulus when abutted by an annular grating of outer diameter 80° (“center+surround”), and two of the annular gratings presented in absence of the center grating (“surround alone”). The lower dashed horizontal line shows the maintained rate in absence of patterned visual stimulus. A SI can be calculated for each annulus direction as the proportional reduction in response from the “center alone” stimulus to the relevant “center+surround” stimulus. Error bars in A, B are ±1 SEM over trials. C, Comparison of tuning for spiking activity (abscissa) and suppression (ordinate). Each unit contributes two points: the open symbols indicate a global measure of direction tuning (Eq. 5) and the filled symbols indicate a similar measure of orientation tuning. D, E, Population averages. D, Population average spiking activity evoked by a large grating, after aligning each neuron to its preferred direction, subtracting the maintained rate, and then normalizing by the mean response across all stimuli. E, Average SI, obtained as in B, after aligning each neuron to the direction of the central grating patch. Dashed horizontal line shows an SI of zero. Error bars in D, E are ±1SEM over neurons. F, Distribution of preferred orientation of suppression, relative to the orientation of the center grating, in units in which the preferred orientation could be defined (orientation tuning index >0.1). A relative orientation of zero indicates neurons in which the most suppressive stimulus was the same orientation as the center; a relative orientation of 90 indicates neurons in which the most suppressive stimulus was orthogonal to the central stimulus. Schematics in panels A, D, E are not to scale.

Impact of adaptation on tuning of spiking activity and surround suppression. A–D, Responses in early (first 0.5 s) time points. A, Population average tuning of spiking activity evoked by a large grating. Conventions as in Figure 4D. B, Population average tuning of suppression induced by annular grating. Conventions as in Figure 4E. C, Comparison of SI for annular gratings that match the direction of the central grating (“parallel,” abscissa), or are tilted by 45° (ordinate). SI for the latter was averaged across both possible directions of tilt. Dashed lines are the unity line. D, Distribution of the difference in suppression for the two annular gratings. Positive indices indicate more suppression by a parallel annulus than a tilted annulus. E–H, Same A–D, but for late (last 0.5 s) time points. Schematics in panels A–C are not to scale.

Adaptation’s effects magnify flexible surround suppression. A–D, Responses in early (first 0.5 s) time points. A, Spiking activity of an example neuron during presentation of a central stimulus either presented alone (C1, C2), or abutting an annulus of varying direction. Two measurements were made, first with a central grating near the preferred direction of spiking activity (C1) and then with a central grating tilted by 45° (C2). Error bars are ±1 SEM over trials. B, Population average tuning of suppression for each of two central gratings (C1, C2). Conventions as in Figure 4E. C, Comparison of suppression induced by pairs of stimuli. The abscissa shows the suppression induced by an annular grating (S1) that matched the direction of the central grating C1. The ordinate shows the suppression induced by an annular grating (S2) that matched the central grating C2. Points below the unity line (dashed line) indicate stronger suppression for S1 than S2. Filled circles show measurements obtained with C1: these generally lie below the unity line, indicating that when C1 is used, S1 is more effective than S2. Open circles show measurements obtained for C2: measurements are generally above the unity line, indicating that S2 is more effective than S1. This is the pattern of results expected if suppression is stronger when the stimuli over the center and surround are matched. D, Distribution of the difference in suppression for each of the center gratings, C1 (lower) and C2 (upper). Positive indices indicate more suppression by S1 than S2. E–H, Same as A–D but for late (last 0.5 s) time points. Example neuron in E is the same as that in A. Schematics above A, E are not to scale.