Article Figures & Data
Figures
- Figure 1.
WM task. On each trial of the task, observers were shown a sample grating in the upper left or right visual field for 6 s. A subsequent change in the color of the fixation point (to red or green; shown here as white) informed observers whether they should remember the sample orientation (memory trials) or simply wait for the next trial to begin (no-memory trials). On memory trials, the cue was followed by a 14 s delay period and the presentation of a 1 s test grating in the same spatial location as the sample. Observers made a two-alternative, forced-choice response indicating whether the sample and test gratings shared the same orientation. On no-memory trials, the 14 s delay period was followed by an additional 1 s blank interval (i.e., no test grating was shown and no response was required). Trials were separated by a 2 s intertrial interval (ITI).
- Figure 2.
Multivoxel pattern analysis of WM data. The results of a multivoxel pattern analysis using the 60 most spatially selective voxels in each V1 ROI during a time period extending from 6 to 16 s after the offset of the sample stimulus. The horizontal line at 0.5 denotes chance classification accuracy. ROIs both contralateral and ipsilateral to the sample stimulus discriminated its orientation but only when active WM maintenance was required. Qualitatively similar results were obtained across a range of pattern sizes (50 and 75 voxels) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material) and also when a linear discriminant algorithm was used to perform classification. Error bars represent ±1 SEM.
- Figure 3.
Event-related analysis of WM data. The mean response of the 60 most spatially selective voxels in V1 (collapsed across hemispheres) is shown as a function of memory condition and ROI. Data are collapsed across stimulus orientation (i.e., 45° vs 135° trials) because no differences in mean response amplitude attributable to this factor were observed. Dashed vertical lines at 0 and 22 s denote the onset of the sample and test stimuli, respectively. Shaded regions represent ±1 SEM.
- Figure 4.
Global orientation-selective patterns are similar during PM and WM maintenance. A classification algorithm was trained to recognize orientation-selective patterns using the 75 most spatially selective voxels from each V1 ROI during PM (both left and right visual field trials were included in this analysis; top panels depict only RVF trials merely for illustrative purposes). As illustrated in the top two panels, this analysis was restricted to conditions in which the sample stimulus appeared in the same cortical ROI during PM and WM maintenance (for an identical analysis that focuses on conditions in which the sample stimulus appears in different spatial locations during each task, see supplemental Fig. 7, available at www.jneurosci.org as supplemental material). This algorithm was then used to infer the orientation of the sample stimulus during each trial of the WM task. The horizontal line at 0.5 denotes chance classification accuracy. Above-chance classification accuracy was observed in ROIs both contralateral and ipsilateral to the sample stimulus, indicating that orientation-selective patterns are similar during PM and WM maintenance. As in Figure 2, we failed to observe sustained above-chance classification accuracy on no-memory trials, indicating that above-chance classification of data from WM trials is not attributable to a lingering sensory effect. Qualitatively similar results were obtained when this analysis was applied to the 65 most spatially selective voxels in each ROI. Error bars represent ±1 SEM.
Tables
- Table 1.
Mean ± SEM discrimination thresholds and accuracy for 45° and 135° trials during PM and WM maintenance
PM task WM task 45° trials 135° trials 45° trials 135° trials Threshold 18.53 ± 1.49 18.40 ± 1.63 9.59 ± 1.26 9.89 ± 0.94 Accuracy 0.74 ± 0.01 0.72 ± 0.02 0.74 ± 0.02 0.74 ± 0.01
Supplemental Data
Files in this Data Supplement:
- supplemental material - Supplemental Material
