The Journal of Neuroscience, April 12, 2006, ():
Estimation of the Timing of Human Visual Perception from Magnetoencephalography
J. Neurosci. Amano et al.
26: 3981
Supplemental data
Files in this Data Supplement:
- supplemental material
-
Supplemental Fig. 1: Model predictions of the stimulus-dependent changes in RT to coherent motion onset from RMS of the MEG response (See also the legend of Fig. 2).
- supplemental material
-
Supplemental Fig. 2: Model predictions of the total variation in RT to coherent motion onset. All hit trials from all motion coherences were divided into five bins of an equal number of trials based on RT, and averaged MEG and median RT were calculated for each bin. The time course of visual response was extracted by using SSP, and all models were applied to predict the change in RT. (See also the legend of Fig. 2.)
- supplemental material
-
Supplemental Fig. 3: The probability that an ideal observer could accurately report hit/miss based on MEG responses at each single trial (Britten et al., 1996). (a,c,e) The distributions of the amplitude of leaky integrated MEG responses (peak amplitude) in hit and miss trials for all (a), 20% (c) and 40% (e) motion coherences. (Responses for 80% motion coherence are not shown since the number of miss trial was too small.) In each panel, the histogram for hit trails is shifted to the right compared with that for miss trials, suggesting that subjects could detect the stimulus when MEG responses were relatively large. (b,d,f) Receiver operating characteristic (ROC), generated by plotting the proportion of hit trials on which leaky integrated MEG responses crossed a threshold against the proportion of miss trials on which integrated responses crossed a threshold, with changing threshold. The circle index on the ROC indicates the performance with the threshold that best accounted for the variation in RT (Fig. 4). (g) Detect probability, defined by the area under the ROC. This index is similar to the choice probability reported in Britten et al., 1996 and indicates the probability that an observer would be able to predict hit/miss correctly when only the amplitude of integrated MEG responses was given. The averaged detect probability was larger than 0.5, and a permutation test within each condition of each subject showed that the detect probability for leaky integrator model was significantly higher than 0.5 for eight out of eight (all), three out of eight (20%), or two out of eight (40%) subjects.
- supplemental material
-
Supplemental Fig. 4: Model predictions of the total variation in RT to chromatic grating onset from the late SSP component of MEG response. (See also the legend of supplemental Fig. 2.)
- supplemental material
-
Supplemental Fig. 5: The results of single-trial analysis of the responses to chromatic grating onset; model predictions of the variation in RT from SSP-extracted visual component. (a,b) The relationship between the actual RTs in each single trial and the detection latencies estimated by the full and leaky integrator models, respectively (subject YO) (See also the legend of Fig. 4). (c) Slope of the regression line between RT and predicted detection latency. (d) Correlation between RT and predicted detection latency. Both measures were significantly larger than zero when the all trials were taken into account.
