Limits...
Adaptive sampling of information in perceptual decision-making.

Cassey TC, Evens DR, Bogacz R, Marshall JA, Ludwig CJ - PLoS ONE (2013)

Bottom Line: Observers compared the direction of two random dot motion patterns that were triggered only when fixated.There were several differences between the optimal observer predictions and human behaviour.These differences point to a number of other factors, beyond the quality of the currently available sources of information, that influences the sampling strategy.

View Article: PubMed Central - PubMed

Affiliation: Department of Computer Science, University of Bristol, Bristol, United Kingdom.

ABSTRACT
In many perceptual and cognitive decision-making problems, humans sample multiple noisy information sources serially, and integrate the sampled information to make an overall decision. We derive the optimal decision procedure for two-alternative choice tasks in which the different options are sampled one at a time, sources vary in the quality of the information they provide, and the available time is fixed. To maximize accuracy, the optimal observer allocates time to sampling different information sources in proportion to their noise levels. We tested human observers in a corresponding perceptual decision-making task. Observers compared the direction of two random dot motion patterns that were triggered only when fixated. Observers allocated more time to the noisier pattern, in a manner that correlated with their sensory uncertainty about the direction of the patterns. There were several differences between the optimal observer predictions and human behaviour. These differences point to a number of other factors, beyond the quality of the currently available sources of information, that influences the sampling strategy.

Show MeSH
The trial sequences of the presented experiments.(a) Comparative two-alternative forced choice decision task. During each trial observers are presented with two stationary RDKs with centres located above and below the centre point of the screen on the vertical axis. The activation of the RDKs was gaze contingent, with a given RDK activated once the observer's gaze fell within the appropriate region of the display. The task was to identify the target pattern: the one whose signal direction was further clockwise through the short angle. RDKs could consist of either black or white dots in both the full and no prior knowledge conditions but the conditions differed in whether the dot polarity was linked to a noise level. A trial started with a preview period of 1000 ms in which observers could process the polarity of the patterns and were instructed by an arrow (appearing 500 ms after trial onset) which pattern to view first. Following the preview, they were free to actively sample the patterns over a period of 1500 ms and subsequently gave their response once both RDKs were extinguished. They were instructed that they must look at each pattern at least once in the course of a trial but other than that, they could switch between patterns in any way they wanted. The figure shows a trial in which the observer was instructed to view the upper pattern first and made only one switch. The eye indicates the vertical eye position, the hatched arrows the signal direction of motion of the moving pattern. The figure is not to scale. (b) Single pattern direction estimation task. Observers viewed a single pattern between 75 and 1000 ms. Once the RDK had offset, they indicated their estimate of the signal direction by moving an arrow to this direction with a mouse and clicking.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3842256&req=5

pone-0078993-g004: The trial sequences of the presented experiments.(a) Comparative two-alternative forced choice decision task. During each trial observers are presented with two stationary RDKs with centres located above and below the centre point of the screen on the vertical axis. The activation of the RDKs was gaze contingent, with a given RDK activated once the observer's gaze fell within the appropriate region of the display. The task was to identify the target pattern: the one whose signal direction was further clockwise through the short angle. RDKs could consist of either black or white dots in both the full and no prior knowledge conditions but the conditions differed in whether the dot polarity was linked to a noise level. A trial started with a preview period of 1000 ms in which observers could process the polarity of the patterns and were instructed by an arrow (appearing 500 ms after trial onset) which pattern to view first. Following the preview, they were free to actively sample the patterns over a period of 1500 ms and subsequently gave their response once both RDKs were extinguished. They were instructed that they must look at each pattern at least once in the course of a trial but other than that, they could switch between patterns in any way they wanted. The figure shows a trial in which the observer was instructed to view the upper pattern first and made only one switch. The eye indicates the vertical eye position, the hatched arrows the signal direction of motion of the moving pattern. The figure is not to scale. (b) Single pattern direction estimation task. Observers viewed a single pattern between 75 and 1000 ms. Once the RDK had offset, they indicated their estimate of the signal direction by moving an arrow to this direction with a mouse and clicking.

Mentions: To assess how humans allocate sampling time in this type of decision problem, we performed an experiment in which observers had to judge which of two RDKs translated in a more clockwise direction. Figure 4 illustrates the paradigm in detail. The comparative nature of this judgement meant that both sources of information had to be inspected in order to come to a decision. The dots moved only when fixated, so that only one pattern could be sampled at a particular point in time. Compared to the standard motion discrimination task with just a single source of evidence, the observer not only had to accumulate evidence from each individual source in order to estimate its direction of motion, but also had to decide at each point in time which information source to sample (i.e., take more samples from the current source or switch to the other source?).


Adaptive sampling of information in perceptual decision-making.

Cassey TC, Evens DR, Bogacz R, Marshall JA, Ludwig CJ - PLoS ONE (2013)

The trial sequences of the presented experiments.(a) Comparative two-alternative forced choice decision task. During each trial observers are presented with two stationary RDKs with centres located above and below the centre point of the screen on the vertical axis. The activation of the RDKs was gaze contingent, with a given RDK activated once the observer's gaze fell within the appropriate region of the display. The task was to identify the target pattern: the one whose signal direction was further clockwise through the short angle. RDKs could consist of either black or white dots in both the full and no prior knowledge conditions but the conditions differed in whether the dot polarity was linked to a noise level. A trial started with a preview period of 1000 ms in which observers could process the polarity of the patterns and were instructed by an arrow (appearing 500 ms after trial onset) which pattern to view first. Following the preview, they were free to actively sample the patterns over a period of 1500 ms and subsequently gave their response once both RDKs were extinguished. They were instructed that they must look at each pattern at least once in the course of a trial but other than that, they could switch between patterns in any way they wanted. The figure shows a trial in which the observer was instructed to view the upper pattern first and made only one switch. The eye indicates the vertical eye position, the hatched arrows the signal direction of motion of the moving pattern. The figure is not to scale. (b) Single pattern direction estimation task. Observers viewed a single pattern between 75 and 1000 ms. Once the RDK had offset, they indicated their estimate of the signal direction by moving an arrow to this direction with a mouse and clicking.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3842256&req=5

pone-0078993-g004: The trial sequences of the presented experiments.(a) Comparative two-alternative forced choice decision task. During each trial observers are presented with two stationary RDKs with centres located above and below the centre point of the screen on the vertical axis. The activation of the RDKs was gaze contingent, with a given RDK activated once the observer's gaze fell within the appropriate region of the display. The task was to identify the target pattern: the one whose signal direction was further clockwise through the short angle. RDKs could consist of either black or white dots in both the full and no prior knowledge conditions but the conditions differed in whether the dot polarity was linked to a noise level. A trial started with a preview period of 1000 ms in which observers could process the polarity of the patterns and were instructed by an arrow (appearing 500 ms after trial onset) which pattern to view first. Following the preview, they were free to actively sample the patterns over a period of 1500 ms and subsequently gave their response once both RDKs were extinguished. They were instructed that they must look at each pattern at least once in the course of a trial but other than that, they could switch between patterns in any way they wanted. The figure shows a trial in which the observer was instructed to view the upper pattern first and made only one switch. The eye indicates the vertical eye position, the hatched arrows the signal direction of motion of the moving pattern. The figure is not to scale. (b) Single pattern direction estimation task. Observers viewed a single pattern between 75 and 1000 ms. Once the RDK had offset, they indicated their estimate of the signal direction by moving an arrow to this direction with a mouse and clicking.
Mentions: To assess how humans allocate sampling time in this type of decision problem, we performed an experiment in which observers had to judge which of two RDKs translated in a more clockwise direction. Figure 4 illustrates the paradigm in detail. The comparative nature of this judgement meant that both sources of information had to be inspected in order to come to a decision. The dots moved only when fixated, so that only one pattern could be sampled at a particular point in time. Compared to the standard motion discrimination task with just a single source of evidence, the observer not only had to accumulate evidence from each individual source in order to estimate its direction of motion, but also had to decide at each point in time which information source to sample (i.e., take more samples from the current source or switch to the other source?).

Bottom Line: Observers compared the direction of two random dot motion patterns that were triggered only when fixated.There were several differences between the optimal observer predictions and human behaviour.These differences point to a number of other factors, beyond the quality of the currently available sources of information, that influences the sampling strategy.

View Article: PubMed Central - PubMed

Affiliation: Department of Computer Science, University of Bristol, Bristol, United Kingdom.

ABSTRACT
In many perceptual and cognitive decision-making problems, humans sample multiple noisy information sources serially, and integrate the sampled information to make an overall decision. We derive the optimal decision procedure for two-alternative choice tasks in which the different options are sampled one at a time, sources vary in the quality of the information they provide, and the available time is fixed. To maximize accuracy, the optimal observer allocates time to sampling different information sources in proportion to their noise levels. We tested human observers in a corresponding perceptual decision-making task. Observers compared the direction of two random dot motion patterns that were triggered only when fixated. Observers allocated more time to the noisier pattern, in a manner that correlated with their sensory uncertainty about the direction of the patterns. There were several differences between the optimal observer predictions and human behaviour. These differences point to a number of other factors, beyond the quality of the currently available sources of information, that influences the sampling strategy.

Show MeSH