Limits...
The Role of Competitive Inhibition and Top-Down Feedback in Binding during Object Recognition.

Wyatte D, Herd S, Mingus B, O'Reilly R - Front Psychol (2012)

Bottom Line: Specifically, we describe how inhibition creates competition among neural populations that code different features, effectively suppressing irrelevant information, and thus minimizing illusory conjunctions.Finally, we argue that temporal synchrony plays only a limited role in binding - it does not simultaneously bind multiple objects, but does aid in creating additional contrast between relevant and irrelevant features.Thus, our overall theory constitutes a solution to the binding problem that relies only on simple neural principles without any binding-specific processes.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychology and Neuroscience, University of Colorado Boulder Boulder, CO, USA.

ABSTRACT
How does the brain bind together visual features that are processed concurrently by different neurons into a unified percept suitable for processes such as object recognition? Here, we describe how simple, commonly accepted principles of neural processing can interact over time to solve the brain's binding problem. We focus on mechanisms of neural inhibition and top-down feedback. Specifically, we describe how inhibition creates competition among neural populations that code different features, effectively suppressing irrelevant information, and thus minimizing illusory conjunctions. Top-down feedback contributes to binding in a similar manner, but by reinforcing relevant features. Together, inhibition and top-down feedback contribute to a competitive environment that ensures only the most appropriate features are bound together. We demonstrate this overall proposal using a biologically realistic neural model of vision that processes features across a hierarchy of interconnected brain areas. Finally, we argue that temporal synchrony plays only a limited role in binding - it does not simultaneously bind multiple objects, but does aid in creating additional contrast between relevant and irrelevant features. Thus, our overall theory constitutes a solution to the binding problem that relies only on simple neural principles without any binding-specific processes.

No MeSH data available.


Related in: MedlinePlus

Neural inhibition in visual binding. We use the LVis model described in O’Reilly et al. (under review) to demonstrate how IT level visual features are suppressed by inhibitory mechanisms over the course of visual processing. tbfTop: Visual occlusion was varied as an independent variable to measure its effect on IT firing patterns during object categorization. Increased occlusion results in a monotonic impairment in categorization accuracy. Bottom: Firing rates were recorded for each IT unit in the model and grouped according to whether they were strongly tuned to the fish category exemplars (dotted lines) or tuned to other categories (solid lines). The first wave of responses from the model’s IT units area code a large number of features, only some of which are category-relevant. Inhibitory competition, however, suppresses the responses of irrelevant non-category units, leaving the features coded by relevant category units to compose the final bound representation. This competitive advantage disappears at higher levels of occlusion (e.g., 50% occlusion) due to fewer category-relevant features being specified in the stimulus.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3376426&req=5

Figure 1: Neural inhibition in visual binding. We use the LVis model described in O’Reilly et al. (under review) to demonstrate how IT level visual features are suppressed by inhibitory mechanisms over the course of visual processing. tbfTop: Visual occlusion was varied as an independent variable to measure its effect on IT firing patterns during object categorization. Increased occlusion results in a monotonic impairment in categorization accuracy. Bottom: Firing rates were recorded for each IT unit in the model and grouped according to whether they were strongly tuned to the fish category exemplars (dotted lines) or tuned to other categories (solid lines). The first wave of responses from the model’s IT units area code a large number of features, only some of which are category-relevant. Inhibitory competition, however, suppresses the responses of irrelevant non-category units, leaving the features coded by relevant category units to compose the final bound representation. This competitive advantage disappears at higher levels of occlusion (e.g., 50% occlusion) due to fewer category-relevant features being specified in the stimulus.

Mentions: In Figure 1, we show the firing patterns of simulated columns of IT neurons when presented with a fish stimulus. Initially, a large number of IT neurons fire, some of which belong to columns that code fish-relevant features and some of which belong to columns that do not. The columns selective to fish-relevant features (e.g., a fish fin, a fish tail), however, quickly out-compete columns selective to fish irrelevant features since the former constitute a better fit with the fish stimulus, increasing their initial evoked response. In turn, the columns selective to fish features inhibit columns selective to irrelevant features, effectively stopping irrelevant neurons from firing and becoming part of the bound representation. Thus, competitive inhibition among detected features helps ensure that a valid combination of features ultimately is bound by driving firing of IT neurons, eliminating invalid conjunctions of features that might lead to binding errors.


The Role of Competitive Inhibition and Top-Down Feedback in Binding during Object Recognition.

Wyatte D, Herd S, Mingus B, O'Reilly R - Front Psychol (2012)

Neural inhibition in visual binding. We use the LVis model described in O’Reilly et al. (under review) to demonstrate how IT level visual features are suppressed by inhibitory mechanisms over the course of visual processing. tbfTop: Visual occlusion was varied as an independent variable to measure its effect on IT firing patterns during object categorization. Increased occlusion results in a monotonic impairment in categorization accuracy. Bottom: Firing rates were recorded for each IT unit in the model and grouped according to whether they were strongly tuned to the fish category exemplars (dotted lines) or tuned to other categories (solid lines). The first wave of responses from the model’s IT units area code a large number of features, only some of which are category-relevant. Inhibitory competition, however, suppresses the responses of irrelevant non-category units, leaving the features coded by relevant category units to compose the final bound representation. This competitive advantage disappears at higher levels of occlusion (e.g., 50% occlusion) due to fewer category-relevant features being specified in the stimulus.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3376426&req=5

Figure 1: Neural inhibition in visual binding. We use the LVis model described in O’Reilly et al. (under review) to demonstrate how IT level visual features are suppressed by inhibitory mechanisms over the course of visual processing. tbfTop: Visual occlusion was varied as an independent variable to measure its effect on IT firing patterns during object categorization. Increased occlusion results in a monotonic impairment in categorization accuracy. Bottom: Firing rates were recorded for each IT unit in the model and grouped according to whether they were strongly tuned to the fish category exemplars (dotted lines) or tuned to other categories (solid lines). The first wave of responses from the model’s IT units area code a large number of features, only some of which are category-relevant. Inhibitory competition, however, suppresses the responses of irrelevant non-category units, leaving the features coded by relevant category units to compose the final bound representation. This competitive advantage disappears at higher levels of occlusion (e.g., 50% occlusion) due to fewer category-relevant features being specified in the stimulus.
Mentions: In Figure 1, we show the firing patterns of simulated columns of IT neurons when presented with a fish stimulus. Initially, a large number of IT neurons fire, some of which belong to columns that code fish-relevant features and some of which belong to columns that do not. The columns selective to fish-relevant features (e.g., a fish fin, a fish tail), however, quickly out-compete columns selective to fish irrelevant features since the former constitute a better fit with the fish stimulus, increasing their initial evoked response. In turn, the columns selective to fish features inhibit columns selective to irrelevant features, effectively stopping irrelevant neurons from firing and becoming part of the bound representation. Thus, competitive inhibition among detected features helps ensure that a valid combination of features ultimately is bound by driving firing of IT neurons, eliminating invalid conjunctions of features that might lead to binding errors.

Bottom Line: Specifically, we describe how inhibition creates competition among neural populations that code different features, effectively suppressing irrelevant information, and thus minimizing illusory conjunctions.Finally, we argue that temporal synchrony plays only a limited role in binding - it does not simultaneously bind multiple objects, but does aid in creating additional contrast between relevant and irrelevant features.Thus, our overall theory constitutes a solution to the binding problem that relies only on simple neural principles without any binding-specific processes.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychology and Neuroscience, University of Colorado Boulder Boulder, CO, USA.

ABSTRACT
How does the brain bind together visual features that are processed concurrently by different neurons into a unified percept suitable for processes such as object recognition? Here, we describe how simple, commonly accepted principles of neural processing can interact over time to solve the brain's binding problem. We focus on mechanisms of neural inhibition and top-down feedback. Specifically, we describe how inhibition creates competition among neural populations that code different features, effectively suppressing irrelevant information, and thus minimizing illusory conjunctions. Top-down feedback contributes to binding in a similar manner, but by reinforcing relevant features. Together, inhibition and top-down feedback contribute to a competitive environment that ensures only the most appropriate features are bound together. We demonstrate this overall proposal using a biologically realistic neural model of vision that processes features across a hierarchy of interconnected brain areas. Finally, we argue that temporal synchrony plays only a limited role in binding - it does not simultaneously bind multiple objects, but does aid in creating additional contrast between relevant and irrelevant features. Thus, our overall theory constitutes a solution to the binding problem that relies only on simple neural principles without any binding-specific processes.

No MeSH data available.


Related in: MedlinePlus