<p>Mean paths of the network’s trajectories during correct (black) and incorrect (red) categorization trials using the easiest short interval (450 ms, <b>A</b>) and the hardest short interval (669 ms, <b>B</b>). It can be appreciated that in “easy” trials (<b>A</b>), incorrect trajectories travel very close to the midline between attractors and remain there for a longer period than incorrect trajectories in “hard” trials (<b>B</b>), which are almost mirror images of the corresponding correct trajectories. This in turn explains the longer RT in incorrect “easy” trials than in “hard” ones.</p
(A) Activity states of an RNN trained on 50k episodes after first three cues. Coloring by first cue ...
Each panel shows data aligned on movement and computed from 1,400 ms before to 400 ms after, as well...
<p>Panel A: mean trajectories for experimental conditions (modified by [<a href="http://www.plosone....
Network activity trajectories on test trials for three levels of training (trained on 0k, 20k, 50k t...
<p>A): Five individual trial trajectories (blue) along with its corresponding hidden target trajecto...
<p>Network activity can be distinguished based on the difference between the ...
A. Neural activity data projected on the PC1 (solid line, linearly detrended) and the correspondent ...
<p>Response of L (<b>A</b>)<b>,</b> S (<b>B</b>), M<sub>L</sub> (<b>C</b>)<b>,</b> M<sub>S</sub> (<b...
<p>Solid lines indicate stimulation during initial rightwards movement; dotted lines show stimulatio...
(A) PCA of network activity after presentation of all cues in a trial for trials answered correctly,...
<p>Trajectories of seen trials (examples shown as green lines) begin in a location distinct from tho...
<p>Left: Representative result from a single trial showing true route computed using noiseless path ...
<p>(<b>A</b>) Topographical plots of the PC coefficients for the first 5 PCs in all 11 subjects, ali...
<p>(<b>A</b>) Illustration of pairwise distances between trajectories within an ensemble (<i>X</i>) ...
<p>To correspond with neural data, only trajectories corresponding to correct decisions are displaye...
(A) Activity states of an RNN trained on 50k episodes after first three cues. Coloring by first cue ...
Each panel shows data aligned on movement and computed from 1,400 ms before to 400 ms after, as well...
<p>Panel A: mean trajectories for experimental conditions (modified by [<a href="http://www.plosone....
Network activity trajectories on test trials for three levels of training (trained on 0k, 20k, 50k t...
<p>A): Five individual trial trajectories (blue) along with its corresponding hidden target trajecto...
<p>Network activity can be distinguished based on the difference between the ...
A. Neural activity data projected on the PC1 (solid line, linearly detrended) and the correspondent ...
<p>Response of L (<b>A</b>)<b>,</b> S (<b>B</b>), M<sub>L</sub> (<b>C</b>)<b>,</b> M<sub>S</sub> (<b...
<p>Solid lines indicate stimulation during initial rightwards movement; dotted lines show stimulatio...
(A) PCA of network activity after presentation of all cues in a trial for trials answered correctly,...
<p>Trajectories of seen trials (examples shown as green lines) begin in a location distinct from tho...
<p>Left: Representative result from a single trial showing true route computed using noiseless path ...
<p>(<b>A</b>) Topographical plots of the PC coefficients for the first 5 PCs in all 11 subjects, ali...
<p>(<b>A</b>) Illustration of pairwise distances between trajectories within an ensemble (<i>X</i>) ...
<p>To correspond with neural data, only trajectories corresponding to correct decisions are displaye...
(A) Activity states of an RNN trained on 50k episodes after first three cues. Coloring by first cue ...
Each panel shows data aligned on movement and computed from 1,400 ms before to 400 ms after, as well...
<p>Panel A: mean trajectories for experimental conditions (modified by [<a href="http://www.plosone....