Limits...
Converging neuronal activity in inferior temporal cortex during the classification of morphed stimuli.

Akrami A, Liu Y, Treves A, Jagadeesh B - Cereb. Cortex (2008)

Bottom Line: IT neurons were selected that responded more strongly to one of the 2 familiar images (the effective image).This sustained component remained linearly dependent on morph level for stimuli more similar to the ineffective image but progressively converged to a single response profile, independent of morph level, for stimuli more similar to the effective image.Computational models suggest that these dynamics could be produced by attractor states and firing rate adaptation within the population of IT neurons.

View Article: PubMed Central - PubMed

Affiliation: SISSA International School for Advanced Studies, Trieste, Italy.

ABSTRACT
How does the brain dynamically convert incoming sensory data into a representation useful for classification? Neurons in inferior temporal (IT) cortex are selective for complex visual stimuli, but their response dynamics during perceptual classification is not well understood. We studied IT dynamics in monkeys performing a classification task. The monkeys were shown visual stimuli that were morphed (interpolated) between pairs of familiar images. Their ability to classify the morphed images depended systematically on the degree of morph. IT neurons were selected that responded more strongly to one of the 2 familiar images (the effective image). The responses tended to peak approximately 120 ms following stimulus onset with an amplitude that depended almost linearly on the degree of morph. The responses then declined, but remained above baseline for several hundred ms. This sustained component remained linearly dependent on morph level for stimuli more similar to the ineffective image but progressively converged to a single response profile, independent of morph level, for stimuli more similar to the effective image. Thus, these neurons represented the dynamic conversion of graded sensory information into a task-relevant classification. Computational models suggest that these dynamics could be produced by attractor states and firing rate adaptation within the population of IT neurons.

Show MeSH
(a) Classification task. After the monkey achieved fixation on a fixation point, a sample, chosen at random among the 9 morphed images or the pair of photographs from which the morphs were made, was presented for 320 ms. Then, after a delay, the photographs appeared together as possible choices (targets). The monkey's task was to pick the target choice that more closely resembled the sample, and make a saccade to it. (b) Behavioral performance. The data are plotted as the proportion of times the monkey chose one of the images (the “effective” image for the cell (see Methods), or Eff) of the 2 original photographs, as a function of the different samples. The trend is linear in the central region between morphs 2 and 8, but performance levels off at the extremes and their nearest neighbors, images Ineff (0) to 2 and 8 to Eff (10). The data are fit with a sigmoid (blue line) and a line (black line). Error bars are standard errors of the mean across different sessions. Images are examples used in one session, where the giraffe was the Ineff image, and the horse the Eff image.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2651479&req=5

fig1: (a) Classification task. After the monkey achieved fixation on a fixation point, a sample, chosen at random among the 9 morphed images or the pair of photographs from which the morphs were made, was presented for 320 ms. Then, after a delay, the photographs appeared together as possible choices (targets). The monkey's task was to pick the target choice that more closely resembled the sample, and make a saccade to it. (b) Behavioral performance. The data are plotted as the proportion of times the monkey chose one of the images (the “effective” image for the cell (see Methods), or Eff) of the 2 original photographs, as a function of the different samples. The trend is linear in the central region between morphs 2 and 8, but performance levels off at the extremes and their nearest neighbors, images Ineff (0) to 2 and 8 to Eff (10). The data are fit with a sigmoid (blue line) and a line (black line). Error bars are standard errors of the mean across different sessions. Images are examples used in one session, where the giraffe was the Ineff image, and the horse the Eff image.

Mentions: On each day, the monkey performed the 2AFC-DMS task (Liu and Jagadeesh 2008) with 2 sample images and 9 morph variants of those images. In each trial, one sample image (or one of its morph variants) was presented, followed by a delay and then followed by a pair of choice images (“choice array”). The monkey's task was to saccade to the image in the choice that most resembled the sample image. An example image pair and associated trials are illustrated in Figure 1. In each trial a red fixation spot (0.3° x 0.3°) appeared at the center of the monitor, and was the cue for the trial to begin. After the monkey acquired fixation, there was a variable delay (250–500 ms) before the onset of the sample image. The sample was presented for 320 ms. After a delay period (700–1100 ms), the choice array (which consisted of both sample images from which the morph variants were created, the Eff and Ineff image) was presented. The choice images were presented 5° up (or down) and to the left of the fixation spot. Location of individual choice images was randomized between the 2 positions (up and down), so the monkey could not determine the location of correct saccade before choice array onset. The different morph variants were presented as samples in random order, until 5–17 trials were recorded for each image.


Converging neuronal activity in inferior temporal cortex during the classification of morphed stimuli.

Akrami A, Liu Y, Treves A, Jagadeesh B - Cereb. Cortex (2008)

(a) Classification task. After the monkey achieved fixation on a fixation point, a sample, chosen at random among the 9 morphed images or the pair of photographs from which the morphs were made, was presented for 320 ms. Then, after a delay, the photographs appeared together as possible choices (targets). The monkey's task was to pick the target choice that more closely resembled the sample, and make a saccade to it. (b) Behavioral performance. The data are plotted as the proportion of times the monkey chose one of the images (the “effective” image for the cell (see Methods), or Eff) of the 2 original photographs, as a function of the different samples. The trend is linear in the central region between morphs 2 and 8, but performance levels off at the extremes and their nearest neighbors, images Ineff (0) to 2 and 8 to Eff (10). The data are fit with a sigmoid (blue line) and a line (black line). Error bars are standard errors of the mean across different sessions. Images are examples used in one session, where the giraffe was the Ineff image, and the horse the Eff image.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2651479&req=5

fig1: (a) Classification task. After the monkey achieved fixation on a fixation point, a sample, chosen at random among the 9 morphed images or the pair of photographs from which the morphs were made, was presented for 320 ms. Then, after a delay, the photographs appeared together as possible choices (targets). The monkey's task was to pick the target choice that more closely resembled the sample, and make a saccade to it. (b) Behavioral performance. The data are plotted as the proportion of times the monkey chose one of the images (the “effective” image for the cell (see Methods), or Eff) of the 2 original photographs, as a function of the different samples. The trend is linear in the central region between morphs 2 and 8, but performance levels off at the extremes and their nearest neighbors, images Ineff (0) to 2 and 8 to Eff (10). The data are fit with a sigmoid (blue line) and a line (black line). Error bars are standard errors of the mean across different sessions. Images are examples used in one session, where the giraffe was the Ineff image, and the horse the Eff image.
Mentions: On each day, the monkey performed the 2AFC-DMS task (Liu and Jagadeesh 2008) with 2 sample images and 9 morph variants of those images. In each trial, one sample image (or one of its morph variants) was presented, followed by a delay and then followed by a pair of choice images (“choice array”). The monkey's task was to saccade to the image in the choice that most resembled the sample image. An example image pair and associated trials are illustrated in Figure 1. In each trial a red fixation spot (0.3° x 0.3°) appeared at the center of the monitor, and was the cue for the trial to begin. After the monkey acquired fixation, there was a variable delay (250–500 ms) before the onset of the sample image. The sample was presented for 320 ms. After a delay period (700–1100 ms), the choice array (which consisted of both sample images from which the morph variants were created, the Eff and Ineff image) was presented. The choice images were presented 5° up (or down) and to the left of the fixation spot. Location of individual choice images was randomized between the 2 positions (up and down), so the monkey could not determine the location of correct saccade before choice array onset. The different morph variants were presented as samples in random order, until 5–17 trials were recorded for each image.

Bottom Line: IT neurons were selected that responded more strongly to one of the 2 familiar images (the effective image).This sustained component remained linearly dependent on morph level for stimuli more similar to the ineffective image but progressively converged to a single response profile, independent of morph level, for stimuli more similar to the effective image.Computational models suggest that these dynamics could be produced by attractor states and firing rate adaptation within the population of IT neurons.

View Article: PubMed Central - PubMed

Affiliation: SISSA International School for Advanced Studies, Trieste, Italy.

ABSTRACT
How does the brain dynamically convert incoming sensory data into a representation useful for classification? Neurons in inferior temporal (IT) cortex are selective for complex visual stimuli, but their response dynamics during perceptual classification is not well understood. We studied IT dynamics in monkeys performing a classification task. The monkeys were shown visual stimuli that were morphed (interpolated) between pairs of familiar images. Their ability to classify the morphed images depended systematically on the degree of morph. IT neurons were selected that responded more strongly to one of the 2 familiar images (the effective image). The responses tended to peak approximately 120 ms following stimulus onset with an amplitude that depended almost linearly on the degree of morph. The responses then declined, but remained above baseline for several hundred ms. This sustained component remained linearly dependent on morph level for stimuli more similar to the ineffective image but progressively converged to a single response profile, independent of morph level, for stimuli more similar to the effective image. Thus, these neurons represented the dynamic conversion of graded sensory information into a task-relevant classification. Computational models suggest that these dynamics could be produced by attractor states and firing rate adaptation within the population of IT neurons.

Show MeSH