Article Figures & Data
Figures
- Figure 1.
V4 localization. A, 3 T structural MR scan of one monkey (right hemisphere, monkey H) with the position of the 19-mm-diameter recording chamber indicated by a white circle. The section is orthogonal to the 55° off-vertical axis of the chamber along which electrode penetrations were made. The plane of section is ∼23 mm from the midline. The lunate sulcus can be seen at the caudal edge of the chamber with the superior temporal sulcus running through the center. B, Left, Ex vivo image of the posterior half of the brain of one monkey (monkey H). Right, Expanded view of a portion of V4 where two magenta dots can be seen. These dots indicate where the electrodes were placed (via dipping electrodes in diiodine) on the final two recordings in monkey H.
- Figure 2.
Behavioral task and trial sequences. A, Order of events in a trial. Monkeys viewed a monitor where a fixation cross appeared at the center. After acquiring fixation and following a variable fixation period, a stimulus array appeared, consisting of a homogenous set of colored distractor disks and a single target of opponent color. Monkeys then were to saccade (denoted with a dashed line) to the target to receive a juice reward. Following reward, there was a short intertrial interval followed by the next trial. B, Example trial sequences used for investigation of priming. Top, Feature priming, sequences of the same target feature (e.g., color, red vs. green) where “1” indicates the first trial where the target took on that feature. Numbers >1 indicate repeated trials of the same target feature. Bottom, Location repetition sequence where 1 indicates that the target appeared in a location different from that of the previous trial. Numbers >1 indicate a repeated target location.
- Figure 3.
Derivation of spiking activity from raw data. Neural activity from each recording site was found by bandpass filtering the raw high-frequency (24,414 Hz) recording, then full wave rectifying, low-pass filtering, and finally converting to z-score and baseline correcting each trial. The results of each step in this procedure are depicted in descending order for a 500 ms window of time taken from a single trial from a session recorded in monkey C. Consequent analysis was then performed on the z-scored data.
- Figure 4.
Priming of pop-out behavioral results. A, Response time to target feature change across both monkeys (n = 2) and all sessions (n = 38). Trial 1 denotes when the target color changed from red to green or vice versa. Trials 5–10 were collapsed as the profile reached asymptote. B, Accuracy as a function of trial since target feature change. Note that accuracy was always above chance (chance = 16.667%, 6 item array).
- Figure 5.
V4 target selection during priming of pop-out. A, Spiking activity aligned on array presentation, averaged across all recording sites (n = 201) when the target was in the receptive field (black) versus when a distractor was in the receptive field (gray). Clouds represent 95% confidence intervals (CIs). Data are broken out by trial since feature change (right) from top to bottom. B, AUC as a function of time from array onset where gray dots are the population average experimental data for each time point, and black curves are the cumulative Weibull fits. Computed target selection times (for details, see Materials and Methods) are indicated with an arrow pointing at the abscissa for each plot.
- Figure 6.
Effect of feature priming on the relationship between behavior and neural processing in V4. A, Response time as a function of target selection time (see Materials and Methods). Dashed vertical and horizontal lines around each data point represent the 95% confidence intervals. The five points plot data from each of the five trials since feature change. Goodness-of-fit through adjusted R2 for the least-squares regression denoted in the bottom right. B, Probability of a correct response as a function of adjusted maximum area under the ROC. Dashed vertical and horizontal lines around each data point represent the 95% confidence interval for probability of correct response and adjusted maximum area under the ROC, respectively. Goodness-of-fit through adjusted R2 for the least-squares regression is denoted in the bottom right.
- Figure 7.
Relationship between neural and behavioral measures for individual units. A, Linear regressions of neural target selection time and behavioral response time for each unit with a measurable change in target selection as a function of feature priming (n = 103). B, Distribution of the slopes plotted in A. The median slope of 0.47, which was significantly different from 0 (black dashed line, ***), is indicated by the red dashed line. C, Linear regressions of adjusted maximum area under the ROC and behavioral accuracy for each of the units (n = 103). D, Distribution of the slopes in C. The median slope was 0.29, which was not significantly different from 0.
- Figure 8.
Target enhancement and distractor suppression during priming of pop-out. A, Comparisons of target response magnitude (left) and distractor response magnitude (right) for trials early in the feature priming sequence (white bars) to trials later in the feature priming sequence (black bars). Error bars denote 2 SEMs. B, Scatter plot showing the difference between target and distractor responses for each recording site. Unprimed activity on the abscissa and primed on the ordinate. Red solid line is the linear regression of the data with 95% CI represented by the red dotted line. Black dashed line indicates the one-to-one line. Values above this line indicate a greater difference between target and distractor responses in primed trials with points below the line indicating a larger magnitude difference in unprimed trials. The regression lies above that line, suggesting a greater target–distractor difference on primed trials.
- Figure 9.
Baseline activity as a function of trial since target feature change. The center of each box is the average activity during the baseline for each condition. Horizontal line through box is the median. Upper and lower limits of the box represent the interquartile range. Whiskers represent the estimated minimum and maximum with dots outside those limits being potential outliers. Activity was measured as the average across all trials since feature change subtracted from each of the conditions. The baseline activity across trials since feature change remains consistent.
- Figure 10.
Behavioral results for target position repetition. A, Response times as a function of trial since target position change across both monkeys (n = 2) and all sessions (n = 38). Trial 1 denotes when the target color changed from red to green or vice versa. Trials 5–10 were collapsed as the profile reached asymptote. B, Accuracy as a function of trial since target position change. Note that accuracy was always above chance (chance = 16.667%, 6-item array).
- Figure 11.
V4 target selection and location-based inhibition of return. A, Spiking activity aligned on array presentation, averaged across all recording sites showing target selectivity (n = 201) when the target is in the receptive field (black) and when a distractor is in the receptive field (gray). Clouds represent 95% confidence intervals. Data are broken out by trial since target position change (right). B, AUC as a function of time from array onset where gray dots are the experimental data and black curves are the cumulative Weibull fits. Computed target selection times (for details, see Materials and Methods) are indicated with an arrow. Data are broken out by trial since target position change (right) from top to bottom.
- Figure 12.
Effect of target position repetition on the relationship between neural responses in V4 and associated behavioral outcome. Response time as a function of target selection time in V4 measured through ROC (for details, see Materials and Methods) for trials where the target location was not a repetition and when it was. The two data points are the trial since target position change groupings from Figure 11. Dashed vertical and horizontal lines denote the 95% CIs around the means.
- Figure 13.
Target suppression with target position repetition. A, Comparisons of target response magnitude (left) and distractor response magnitude (right) for trials where the prior target was in a different location (white bars) to trials where the target location repeats (black bars). Error bars denote 2 SEMs. B, Scatter plot showing the difference between target and distractor responses for each recording site. Nonrepeat target location activity on the abscissa and repeated target location on the ordinate. Red solid line is the linear regression of the data with 95% CIs represented by the red dotted line. Black dashed line indicates the 1-to-1 line. Values above this line indicate a greater difference between target and distractor responses in target position repeat trials, with points below the line indicating a larger magnitude difference in target position change trials. The regression lies below that line, suggesting a greater target–distractor difference on target position change trials.
In this issue
