The mean auditory word length was The pixelwise overlap computed by summing all stimuli images per object, per category the rightmost column , and per shape type the bottom row. There were eight different objects and 40 unique object images. The objects comprised of two categories and four shape types. The object similarity judgements for both the category and shape dimensions were collected from all participants.
Participants completed the rating task prior to the EEG task. During the rating task they were presented with a word and an array of objects, and were asked to indicate how similar each object is to the word, on a scale from 1 to 5. Each participant rated all 40 images relative to all eight words. The rating procedure was repeated twice, once for the shape dimension and once for the category dimension Fig. Similarity matrices and reaction times.
Prior to the experiment, participants completed the rating task where they indicated how similar each target object is to each cue word, on a scale from one to five, separately for the shape and for the category dimensions. Panels A and B show the similarity ratings averaged per cue-target pair and across subjects, for the Shape A and Category B dimensions red represents large similarities.
During the main experiment, participants replied with the button press if the target object matched or mismatched the cue word. Panel C shows the reaction times in ms averaged per cue-target pair and across subjects red represents slower reaction times. Note that in all reported analyses we used individual, rather than group-averaged similarity ratings and reaction times.
The group-averaged data are only shown here for illustration. The similarity data A , B and the reaction times C on the diagonals of the matrices, which correspond to the congruent pairs, are not shown, because only incongruent trials were used for the analysis. Participants completed trials of the word-picture matching task. On each trial, participants heard a cue word a fruit or vegetable name , followed by a picture after one second delay.
They were instructed to respond via button press whether the picture matched the word yes or no.
Each incongruent combination of a cue word and a target picture was repeated 12 times, and each congruent pair 36 times. The order of trials was randomised across participants. An equidistant electrode cap was used to position 60 electrodes on the scalp.
EEG data were recorded against the reference at the right mastoid; an additional electrode measured the voltage on the left mastoid, and the data were offline converted to a linked-mastoids reference. Bipolar electrooculogram EOG was computed using electrodes that were placed horizontally and vertically around the eyes. Segments containing eye-movements, or muscle artifacts were identified based on signal variance. Identified segments were inspected visually and rejected if contamination with artifacts was confirmed. On average, 8.
The data were subsequently bandpass filtered from 0. Finally, using independent component analysis, artifacts caused by blinks and other events not related to brain activity were removed from the data. This resulted in a vector of 56 mean RT values per participant. We then used a correlation analysis to test if the RTs are explained by the similarity between cues and pictures. To elaborate, for each participant, we computed a Spearman rank correlation between the word-picture similarity and the corresponding RTs.
We tested two different similarity models: i subjective shape similarity per subject and ii and subjective category similarity per subject. Thus, the averaging resulted in 56 ERP waveforms for each channel. We further used a correlation-based analysis to test if the pattern of the evoked responses across the word-picture pairs could be explained by the similarity between cues and pictures.
We tested two similarity models: i subjective shape similarity per subject and ii subjective category similarity per subject. Correlations were computed for each channel and time point.
This resulted in a channel x time matrix of correlation coefficients for each participant and for each model. A similar analysis approach has been used in a priming experiment before We performed this test separately for each similarity model. At the group level, we aim to test if the correlations in each channel x time sample are different from zero.
However, testing each time x channel sample independently leads to massive multiple comparisons. To account for the multiple comparisons problem, we used nonparametric cluster-based permutation statistics approach In this method, the complete channel x time matrix is tested by computing a single test statistic, and therefore, the multiple comparisons problem is resolved.
We elaborate on this procedure in the following paragraphs. We followed the procedure described previously 45 , We first computed a paired-sample t-test for each channel x time point, where we compared the correlation coefficients from 20 participants with a vector of 20 zeros. All t values above a threshold corresponding to an uncorrected p value of 0. This step was performed separately for samples with positive and negative t values two-tailed test. The t values within each cluster were then summed to produce a cluster-level t score cluster statistic.
The attentional enhancement of the hemodynamic 55 , 56 , 57 and electrophysiological 58 responses in the visual cortex is well known. Alsius, A. Wiley Interdiscip. Detection and resolution of audio-visual incompatibility in the perception of vowels. Routinely fails to observe or recognize changes in bulletin board displays, signs or posted notices. The user must choose the correct answer as quickly as possible.
This statistic was entered in the cluster-based permutation procedure To obtain a randomization distribution to compare with the observed clusters, we randomly exchanged the condition labels between the true and null conditions that is, the vector of zero correlations, same as described above. We then computed the paired sample t-test. This step was repeated across permutations of the data. For each permutation, we computed the cluster-sums of subthreshold t-values.
The most extreme cluster-level t score on each iteration were retained to build a null hypothesis distribution. The position of the original real cluster-level t scores within this null hypothesis distribution indicates how probable such an observation would be if the null hypothesis was true no systematic difference from 0 in correlations across participants. The sensitivity of the cluster-based ERP statistics depends on the length of the time interval that is analysed. To increase the sensitivity of the statistical test, it is therefore recommended to limit the time interval on the basis of prior information about the time course of the effect.
The average shape similarity rating across all cue-target pairs across all subjects was 2. The average category similarity rating across all cue-target pairs across all subjects was 2.
The topoplots on the Fig. Additionally, we obtained several marginally significant clusters. The larger the marker, the longer the channel remained statistically significant within the given interval. Correlation values plotted against the ERPs. ERPs black are averaged over all cue-target combinations over all participants. A selection of 16 channels out of original 60 corresponding to the standard 10—20 electrode system is shown. The similarity in shape between cues and targets affected the entire dynamics of the visual processing.
As shown by the ERP waveforms in Fig. Notably, the spatial and temporal extent of this cluster was similar to that of the latest shape similarity cluster. We hypothesised that this effect could be driven by non-independency between the shape and category similarity ratings. All these pairs kiwi-potato, banana-zucchini, pear-eggplant, apricot-onion, and the respective reversed pairs were different in category.
Thus, the correlation between the shape and category similarity was driven by these pairs. To tease apart the effect of shape from that of category, we ran two additional post-hoc analyses. Second, we repeated the original correlation analysis while excluding the eight word-picture pairs that drove the correlation between shape and category similarity. The results of the partial correlation were similar to the normal correlation, however the magnitude of the late category effect has reduced.
The second post-hoc analysis yielded a similar picture: the late ERP responses still correlated with the word-picture category similarity, but the effect became smaller and dropped below the significance threshold. This indicates, that the late category effect could be, at least partly, explained by the association between the shape and category in the designed stimuli. The results of the partial correlation analysis of the shape similarity did not differ from the results of the main analysis and are not shown in the table.
Linguistic labels are known to facilitate object recognition, yet the mechanism of this facilitation is not fully understood.
Differences in Visual Perception: The Individual Eye examines the differences in visual perception that can occur in various circumstances when observers. Individual differences. Vision. Visual mechanisms. Experimental design .. of ocular tracking were driven by different mechanisms of motion perception.
A large number of psychophysical studies have suggested that words activate the visual representation of their reference, and particularly its most salient features, such as visual shape 17 , 19 , In the present work we aimed to tease apart the visual shape and the semantic category effects of words on object recognition, and study the dynamics of these effects at the neural level. We conducted an EEG word-picture matching experiment, using objects from two categories and with four different shapes. Contrary to our expectations, we found that only the word-picture shape similarity, but not the category similarity robustly predicts the reaction times.
Here we have extended this earlier finding by showing, unambiguously, that this early effect on visual processing can reflect an anticipation of the upcoming visual object shape.