Limits...
Training top-down attention improves performance on a triple-conjunction search task.

Baluch F, Baluchg F, Itti L - PLoS ONE (2010)

Bottom Line: Training has been shown to improve perceptual performance on limited sets of stimuli.However, whether training can generally improve top-down biasing of visual search in a target-nonspecific manner remains unknown.Subjects became experts at this task, with twofold increased performance, decreased fixation duration, and stronger tendency to guide gaze toward items with color and spatial frequency (but not necessarily orientation) that resembled the target, suggesting improved general top-down biasing of search.

View Article: PubMed Central - PubMed

Affiliation: Neuroscience Graduate Program, University of Southern California, Los Angeles, California, USA.

ABSTRACT
Training has been shown to improve perceptual performance on limited sets of stimuli. However, whether training can generally improve top-down biasing of visual search in a target-nonspecific manner remains unknown. We trained subjects over ten days on a visual search task, challenging them with a novel target (top-down goal) on every trial, while bottom-up uncertainty (distribution of distractors) remained constant. We analyzed the changes in saccade statistics and visual behavior over the course of training by recording eye movements as subjects performed the task. Subjects became experts at this task, with twofold increased performance, decreased fixation duration, and stronger tendency to guide gaze toward items with color and spatial frequency (but not necessarily orientation) that resembled the target, suggesting improved general top-down biasing of search.

Show MeSH

Related in: MedlinePlus

Saccade statistics.(a) Main sequence, plotting saccade amplitudes against peak velocity for the first session (red) and fifth session (blue). Overlap shows no difference in main sequence. (b) Intersaccadic interval reduces with session data. Points were computed by pooling saccades for each session for all subjects and taking a mean. Error bars are SEM. (c) Reaction time as a function of number of saccades. Regression line shows significant correlation (). (d) Reaction time as a function of intersaccadic interval. Regression shows weak correlation ().
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2823769&req=5

pone-0009127-g004: Saccade statistics.(a) Main sequence, plotting saccade amplitudes against peak velocity for the first session (red) and fifth session (blue). Overlap shows no difference in main sequence. (b) Intersaccadic interval reduces with session data. Points were computed by pooling saccades for each session for all subjects and taking a mean. Error bars are SEM. (c) Reaction time as a function of number of saccades. Regression line shows significant correlation (). (d) Reaction time as a function of intersaccadic interval. Regression shows weak correlation ().

Mentions: The eye movements of all the subjects were grouped by session, and statistics were then computed on this data. We first analyzed the main sequence, which plots peak velocity against saccadic amplitude. The main sequences for session one and session five are shown in figure 4a. To determine whether there was a difference between the two sequences we first fitted a linear function to the main sequence of session one and then used this model to predict saccade amplitudes using the peak velocity data from session five saccades. We then ran a two-sample t-test between predicted saccade amplitudes and real saccade amplitudes for session five and found no significant difference (p = 0.50). The analysis of the main sequences therefore revealed no effect of training on these saccade statistics, and the subjects’ eye movements were similar in this regard. Similarly, no significant trend was found in saccadic amplitude or velocity individually (data not shown). However, when we analyzed the ISI we found a significant drop from early sessions in training to late sessions, as illustrated in figure 4b. Specifically, a one-way ANOVA showed a strong effect (F(9,73481) = 43.95, ) of session on intersaccadic interval. These results demonstrate a change in saccadic strategy on the part of the observes, a change marked by increased efficiency in examining the Gabor patches and greater speed in rejecting non-target Gabor patches. As expected a fall in ISI resulted in a drop in reaction time (RT). However, we found that RT was more strongly dependent on the number of saccades made rather than on ISI. We found a significant dependence () of RT on the number of saccades made (figure 4c). A weaker dependence (figure 4d) of RT on ISI was found (). The data shown in the figures is for trials where reaction time was ; the results for the full dataset were similar (RT vs saccade count and RT vs ISI ). Therefore number of saccades appeared to be more important in determining RT than ISI.


Training top-down attention improves performance on a triple-conjunction search task.

Baluch F, Baluchg F, Itti L - PLoS ONE (2010)

Saccade statistics.(a) Main sequence, plotting saccade amplitudes against peak velocity for the first session (red) and fifth session (blue). Overlap shows no difference in main sequence. (b) Intersaccadic interval reduces with session data. Points were computed by pooling saccades for each session for all subjects and taking a mean. Error bars are SEM. (c) Reaction time as a function of number of saccades. Regression line shows significant correlation (). (d) Reaction time as a function of intersaccadic interval. Regression shows weak correlation ().
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2823769&req=5

pone-0009127-g004: Saccade statistics.(a) Main sequence, plotting saccade amplitudes against peak velocity for the first session (red) and fifth session (blue). Overlap shows no difference in main sequence. (b) Intersaccadic interval reduces with session data. Points were computed by pooling saccades for each session for all subjects and taking a mean. Error bars are SEM. (c) Reaction time as a function of number of saccades. Regression line shows significant correlation (). (d) Reaction time as a function of intersaccadic interval. Regression shows weak correlation ().
Mentions: The eye movements of all the subjects were grouped by session, and statistics were then computed on this data. We first analyzed the main sequence, which plots peak velocity against saccadic amplitude. The main sequences for session one and session five are shown in figure 4a. To determine whether there was a difference between the two sequences we first fitted a linear function to the main sequence of session one and then used this model to predict saccade amplitudes using the peak velocity data from session five saccades. We then ran a two-sample t-test between predicted saccade amplitudes and real saccade amplitudes for session five and found no significant difference (p = 0.50). The analysis of the main sequences therefore revealed no effect of training on these saccade statistics, and the subjects’ eye movements were similar in this regard. Similarly, no significant trend was found in saccadic amplitude or velocity individually (data not shown). However, when we analyzed the ISI we found a significant drop from early sessions in training to late sessions, as illustrated in figure 4b. Specifically, a one-way ANOVA showed a strong effect (F(9,73481) = 43.95, ) of session on intersaccadic interval. These results demonstrate a change in saccadic strategy on the part of the observes, a change marked by increased efficiency in examining the Gabor patches and greater speed in rejecting non-target Gabor patches. As expected a fall in ISI resulted in a drop in reaction time (RT). However, we found that RT was more strongly dependent on the number of saccades made rather than on ISI. We found a significant dependence () of RT on the number of saccades made (figure 4c). A weaker dependence (figure 4d) of RT on ISI was found (). The data shown in the figures is for trials where reaction time was ; the results for the full dataset were similar (RT vs saccade count and RT vs ISI ). Therefore number of saccades appeared to be more important in determining RT than ISI.

Bottom Line: Training has been shown to improve perceptual performance on limited sets of stimuli.However, whether training can generally improve top-down biasing of visual search in a target-nonspecific manner remains unknown.Subjects became experts at this task, with twofold increased performance, decreased fixation duration, and stronger tendency to guide gaze toward items with color and spatial frequency (but not necessarily orientation) that resembled the target, suggesting improved general top-down biasing of search.

View Article: PubMed Central - PubMed

Affiliation: Neuroscience Graduate Program, University of Southern California, Los Angeles, California, USA.

ABSTRACT
Training has been shown to improve perceptual performance on limited sets of stimuli. However, whether training can generally improve top-down biasing of visual search in a target-nonspecific manner remains unknown. We trained subjects over ten days on a visual search task, challenging them with a novel target (top-down goal) on every trial, while bottom-up uncertainty (distribution of distractors) remained constant. We analyzed the changes in saccade statistics and visual behavior over the course of training by recording eye movements as subjects performed the task. Subjects became experts at this task, with twofold increased performance, decreased fixation duration, and stronger tendency to guide gaze toward items with color and spatial frequency (but not necessarily orientation) that resembled the target, suggesting improved general top-down biasing of search.

Show MeSH
Related in: MedlinePlus