Skip to main content
. Author manuscript; available in PMC: 2009 Apr 1.
Published in final edited form as: J Neurosci. 2008 Oct 1;28(40):10045–10055. doi: 10.1523/JNEUROSCI.2142-08.2008

Figure 4.

Figure 4

IT selectivity and behavioral performance within object sets following training. a) Fraction of IT neurons with responses that were selective among the “restricted-experience” objects when those objects were tested at the trained retinal position (black bars) and at the non-trained, equally eccentric position (white bars). Data from both monkeys have been combined (see Fig. 2). As in Figure 2, selectivity was determined by ANOVA and the x-axis shows a range of significance levels (p values) used for deeming a neuron to be “selective”. Asterisks indicate significant difference at the trained and non-trained positions (p<0.05, chi-square test). Portions of the bars overlaid in gray indicate the fraction of neurons that were selective at both the trained and non-trained positions. b) Left side: Performance of the population of selective IT neurons (p < 0.1, ANOVA), assessed by linear classification (see Methods), shown separately for each monkey. Chance is 25% (four possible objects). Double asterisks indicate significant performance difference at the trained and non-trained positions (p<0.01, test of independent proportions). Right side: behavioral performance for one monkey at discriminating among the “restricted-experience” objects at the trained (black) and non-trained (white) retinal positions (with interleaved low-contrast trials from the “bias test” object set; see Methods). While absolute behavioral performance (panels b and d, left side) is decreased from levels achieved during training because several trial types are interleaved, a strong asymmetry is observed between the trained and non-trained positions for the “restricted-experience” objects, but not the interleaved “bias test” objects. Double asterisks indicate significant difference (p<0.01, test of independent proportions). c,d) Same conventions as (a,b) for the “bias test” objects, which show no asymmetry between the trained and untrained retinal positions [please note that no direct comparison is implied between the absolute selectivity/performance levels for the “restricted experience” (a,b) and “bias test” objects (c,d) ].