Fig 1.
Stimuli and experimental paradigms.
A) Stimulus space described by two parameters derived from the distribution of local contrast, contrast energy (CE) and spatial coherence (SC). In this space, simple images containing one or a few easily segmentable objects are on the lower left, while complex images with a high degree of fragmentation are on the upper right. Thumbnails show 100 images (50 animal, 50 non-animal) randomly drawn from the larger image set from which the stimuli were selected. B) Image statistics of the stimuli: each point represents a scene sampled from the image space described in A). Scenes had either low (red), medium (green) or high (blue) CE and SC values. Within these conditions, CE and SC values were matched between scenes with target objects (animals: “A”, filled dots) and without target objects (non-animals: “NA”, open dots). C) Exemplars from each stimulus complexity condition. D) Experimental design of Experiment 1 (fMRI). On GO trials, subjects indicated whether the scene contained a target object or not. On STOP trials, an auditory signal followed stimulus presentation after a variable inter-trial-interval (ITI), signaling that subjects had to withhold their response. Only GO trials were analyzed. E) Experimental design of Experiment 2 (EEG). Participants were instructed to respond as fast as possible on speed trials and as accurate as possible on accuracy trials and received feedback on each trial.
Fig 2.
Behavioral results of the fMRI experiment (Experiment 1).
A) Average reaction times (RT) for the animal/non-animal categorization task per condition. B) Accuracy (percentage correct) per condition. Horizontal black lines indicate the statistical outcome of one-way repeated-measures ANOVAs; gray lines indicate the results of pairwise tests corrected for multiple comparisons using a Sidák correction. Error bars represent S.E.M. * = p < 0.05, # = p < 0.10.
Fig 3.
A) Statistical parametric maps for the animal (A) > non-animal (NA) contrast for each condition. From left to right, MNI coordinates for each transversal slice are z = [10, 2, –6]. B) Statistical parametric maps for the non-animal > animal contrast for each condition. From left to right, MNI coordinates are z = [–2, –8, –14]. C) Differences in the differential animal vs. non-animal activity between conditions. MNI-coordinates: [x = 6, y = -88, z = 6]. Maps were cluster-corrected and thresholded at z = 2.3. Color scales range from z = 2.5 to 5 (A and B) and z = 2.3 to 3.5 (C).
Table 1.
Whole brain fMRI analysis cluster coordinates for the significant contrasts.
COG = center of gravity, p = p-value, lat. = lateral, occ. = occipital, temp. = temporal, inf. = inferior, post. = posterior, L = left, R = right. Coordinates are reported in MNI space. Areas included in the clusters were determined by determining overlap of local maxima within clusters with the probability maps of the Juelich Histological Atlas and the Harvard-Oxford Cortical Structural Atlas implemented in FSL. Note that for the contrasts HIGH (animal—non-animal), MED (non-animal—animal) and [HIGH (animal-non-animal)—MED (animal- non-animal)], the clusters are bilateral, forming one cluster across hemispheres (see also Fig 3).
Fig 4.
Differential activity for animal vs. non-animal scenes for each condition in A) V1, B) PPA, C) LOC and D) FFA. Horizontal black lines indicate the statistical outcome of repeated-measures ANOVAs; gray lines indicate the results of pairwise tests corrected for multiple comparisons using a Sidák correction. * = p < 0.05, # = p < 0.10. Error bars represent S.E.M.
Fig 5.
Behavioral and HDDM modeling results of the EEG experiment (Experiment 2).
A) Average reaction times (RT) per condition and task instruction. B) Accuracy (percentage correct) per condition and task instruction. C) Estimates of drift rates per condition, indicating slower evidence accumulation for high complexity scenes. D) Estimates of the amount of evidence required (boundary) per condition and task instruction. Gray lines indicate the results of pairwise post-hoc tests between conditions corrected for multiple comparisons using a Sidák correction. Error bars represent S.E.M. * = p < 0.05.
Fig 6.
ERP results from the EEG experiment (Experiment 2).
Average ERP amplitude for animal (solid lines) and non-animal scenes (dashed lines) per condition for an occipital-peri-occipital cluster of EEG channels (Oz, POz, O1, O2, PO3, PO4, PO7, PO8) for the A) speed and B) accuracy trials. Shaded regions indicate SEM across participants. Significant differences between animal and non-animal ERPs (FDR-corrected) are indicated with thick black lines (also in C/D). C-D Animal vs. non-animal difference wave for each condition overlaid and statistically compared (FDR-corrected) for the speed A) and accuracy B) trials. Thick lines at the bottom of the graphs indicate significant difference wave deflections from zero, and shadings indicate S.E.M. Symbol markers indicate significant differences in difference wave amplitude between conditions, blue asterisk: HIGH vs. LOW; blue plus: HIGH vs. MEDIUM; green plus: MEDIUM vs. LOW.
Fig 7.
Posterior probabilities of the regression weight for drift rate (v) on the single-trial ERP amplitude averaged between 220–325 ms after stimulus onset for LOW, MED and HIGH complexity.