Figure - PMC

Skip to main content

An official website of the United States government

Here's how you know

Here's how you know

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

View full-text article in PMC

. 2017 Dec 23;6:e29226. doi: 10.7554/eLife.29226

Search in PMC
Search in PubMed
View in NLM Catalog
Add to search

© 2017, Gutnisky et al

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

PMC Copyright notice

Figure 5. — (A) The probability of correct classification (PCC) as a function of population size. PCC was significantly higher in the low pre-stimulus case (F(1,164)=9.32; p<0.005; two-way repeated measures ANOVA). (inset) The difference in classification performance between low and high pre-stimulus response states for a small (n = 2, orange) and a large population (n = 12, blue). The performance difference was greater for the larger neural population (p<0.05, bootstrap test). (B) Noise correlations of evoked responses for the high and low pre-stimulus states – correlations were significantly higher in the high pre-stimulus state (p=7.918e-10; paired t-test). (C) Probability of correct classification for the two cortical states. ‘With correlations’ represents data using the following equation: $d^{2} = Δ μ^{T} Q^{- 1} Δ μ$ ; Probability of correct classification = $e r f c (- \sqrt{d^{2}}) / 2$ . ‘Without correlations’ represents the probability of correct classification when ignoring the effect of noise correlations using ( $d_{s h u f f l e d}^{2} = Δ μ^{T} Q_{d}^{- 1} Δ μ$ ), where $Q_{d}$ is the diagonal covariance matrix. In each condition there is a statistically significant difference between the high and low pre-stimulus conditions (*p<0.05; paired t-test). (D) The magnitude of the difference in FLD means (left) and the magnitude of the pooled standard deviation ( $σ$ ) of the FLD (right). In the low pre-stimulus condition, the difference in means was significantly greater (p<0.0001; paired t-test). The average variance was also higher in the low pre-stimulus condition (p<0.0001; paired t-test), but this had less overall impact on the population d’. The results in this figure were obtained for all the cells recorded across sessions for an orientation difference of ±5°.

Figure 5—figure supplement 1. — (A) The probability of correct classification (PCC) as a function of population size. PCC was significantly higher in the low pre-stimulus case (F(1,164)=9.32; p<0.005; two-way repeated measures ANOVA). (inset) The difference in classification performance between low and high pre-stimulus response states for a small (n = 2, orange) and a large population (n = 12, blue). The performance difference was greater for the larger neural population (p<0.05, bootstrap test). (B) Noise correlations of evoked responses for the high and low pre-stimulus states – correlations were significantly higher in the high pre-stimulus state (p=7.918e-10; paired t-test). (C) Probability of correct classification for the two cortical states. ‘With correlations’ represents data using the following equation: $d^{2} = Δ μ^{T} Q^{- 1} Δ μ$ ; Probability of correct classification = $e r f c (- \sqrt{d^{2}}) / 2$ . ‘Without correlations’ represents the probability of correct classification when ignoring the effect of noise correlations using ( $d_{s h u f f l e d}^{2} = Δ μ^{T} Q_{d}^{- 1} Δ μ$ ), where $Q_{d}$ is the diagonal covariance matrix. In each condition there is a statistically significant difference between the high and low pre-stimulus conditions (*p<0.05; paired t-test). (D) The magnitude of the difference in FLD means (left) and the magnitude of the pooled standard deviation ( $σ$ ) of the FLD (right). In the low pre-stimulus condition, the difference in means was significantly greater (p<0.0001; paired t-test). The average variance was also higher in the low pre-stimulus condition (p<0.0001; paired t-test), but this had less overall impact on the population d’. The results in this figure were obtained for all the cells recorded across sessions for an orientation difference of ±5°.