In summary, we find that low-level image features drove

In summary, we find that low-level image features drove GDC-0199 chemical structure the fixations performed

by the monkeys that actively explore the natural scenes if the images did not show faces of primates. For the remaining images, most of the eye movements relate to faces, i.e., regions that are typically of low saliency value and thus have a low bottom–up impact. Our analysis of the fixation positions (Section 2.1) revealed that these are not evenly distributed across the images, but rather tend to occur clustered in space (Fig. 3). Our interpretation was that these clusters represent ROIs. Thus, our next aim is to gain insight on the temporal sequences of visiting these ROIs. Therefore we explored the scanpaths of the image explorations by applying a Markov chain (MC) analysis to the eye movement trajectories

(see details in Section 4.5). Thereby we assume each of the significantly identified Stem Cell Compound Library screening fixation clusters on a particular image as a Markov state, and estimate the probabilities for consecutive fixations to either stay in the same cluster, to switch to a different cluster, or end up in the background. In this analysis the assumption of a MC enters in that the next state will be reached only depending on the current state, but does not depend on past states (see details in Section 4.5). The cluster analysis of the fixation positions typically revealed 3 to 5 significant clusters per image for monkeys D and M, however, not a single significant cluster could be extracted for monkey S. Thus this monkey seems not to express subjective ROIs, and we had to conclude that this monkey is not actively exploring the images. Since the MC analysis is based on ROIs, monkey S had to be excluded from the subsequent analysis of the sequence of fixation positions. Fig. 5A shows examples of eye movement sequences (4 out of 33) of monkey

D during presentations of the same image. The fixation positions L-gulonolactone oxidase of monkey D on the image during all its presentations were grouped into three significant clusters (Fig. 5B, color coded). Fixation positions that do not belong to any identified cluster (small blue dots) are pooled together and assigned to the background cluster (see Sections 4.6 and 4.7). The result of the MC analysis on these data is shown in Fig. 5C as a transition graph. Each identified significant cluster, as well as the background cluster, represents a state of the model, whereas the transitions between the states (whose probabilities are indicated in black) are marked by directed arrows. The statistical significance was evaluated by comparing the transition probabilities of the empirical data to uniform probabilities (Fig. 5C, numbers in gray; details see Section 4.7). The probabilities (across all images) of staying within the significant clusters are 87% (40/46) for monkey D and 95% (19/20) for monkey M, thus significantly higher than expected by chance (Fig. 5D). In contrast, the probabilities of moving between significant clusters (Fig.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>