Subcortical Mechanisms of Feature-Based Attention

Detection probability. The probability of detection of an event in experiment 2 at a given time during the stimulus cycle is plotted as a mean across subjects. During the first 20 s of each cycle, subjects attended to the moving dots and detected changes in the direction or motion coherence. During the second 20 s of each cycle, subjects attended to the colored dots and detected color convergences.

Regions of interest. Two coronal slices are shown from experiment 1 in the same well activated subject, zoomed to show the detail in the subcortex. The second slice is 4 mm posterior to the first. ROIs are shown for the LGN, LP, SC, and IP. The circular color legend in the upper right indicates the locations in the visual field to which voxels of a given color best responded. Left (L), right (R), superior (S), and inferior (I) directions are labeled.

Volumetric distribution of the retinotopic visual field representation. Each point in the polar plots indicates the average volume of each nucleus, across bilaterally activated subjects in experiment 1, representing a 22.5° segment of the visual field at the given polar angle. Abbreviations are as defined in Figure 3; the scales differ among the nuclei. The shaded areas indicate the extent of the SEM. The direction terms orient the plots to locations in the visual field.

Spatial attention. a, The mean response amplitudes across subjects are shown for each nucleus (abbreviations are as defined in Fig. 3) for the moving and colored dot stimuli during the attended (white bars) and unattended (hatched bars) conditions in experiment 1. Error bars indicate the SEM over the subjects. The p value is shown for the two-tailed t test between the attentional conditions for each nucleus and stimulus: *p < 0.05, **p < 0.01, ***p < 0.001. In addition, in the LGN, the mean response to the moving dot fields was significantly greater than that to the colored dot fields. b, The mean AMIs are shown for the moving dot (white bars) and colored dot (shaded bars) fields in each nucleus. The p values indicate deviance from zero. Conventions are as in a.

Experiment 2 mean time series. The mean response over subjects in each nucleus (abbreviations are as defined in Fig. 3) during one stimulus period is shown. The shaded area indicates the extent of the SEM over subjects. During the first half of each stimulus period, the subjects attended to the moving dot field. The hatched region indicates the second half of the stimulus period during which the subjects attended to the colored dot field.

Experiment 2 response modes. The mean time series in each nucleus (abbreviations are as defined in Fig. 3) across all scanning runs and all subjects was correlated with two frequencies: f₀, the fundamental stimulus frequency (motion vs color, white bars), a period of 40 s corresponding to one stimulus cycle between attending to the moving dots and colored dots; and f₁, double the fundamental frequency (attention switching, shaded bars), a period of 20 s corresponding to changes in the attentional state. The dashed lines indicate the significance levels of the correlation coefficients.

Experiment 2 power spectra. Each graph shows the power spectra of the mean response across subjects in each ROI during experiment 2. The open circle marks the power at six cycles/scan, which was the fundamental frequency of the experiment, corresponding to the difference between the attention conditions. The closed circle marks 12 cycles/scan, which was the frequency of the switches between the attentional conditions.

Experiment 2 response phases. The mean response amplitudes over all voxels in all subjects at f₀, the fundamental frequency of the motion vs color attention cycle (red lines), or f₁, the frequency of attentional switches (green lines), are plotted as a function of response phase in polar coordinates in the left two plots and in Cartesian coordinates in the right two plots. The upper two plots are for the LGN and the lower two the SC.

Experiment 2 responses by subject. The gray circles plot for each subject the phases and amplitudes of the responses in experiment 2 as a vector average of the responses of all of the individual voxels in each ROI. The black circles plot the vector average over all subjects. The left two plots indicate the responses at the fundamental frequency, f₀, which is the frequency of one motion vs color attention cycle. The right two plots indicate the responses at the first harmonic, f₁, which is the frequency of the switches between attentional states. The upper two plots are for the LGN and the lower two the SC.