Aging Potentiates Lateral but Not Local Inhibition of Orientation Processing in Primary Visual Cortex

Abstract

Aging-related declines in vision can decrease well-being of the elderly. Concerning early sensory changes as in the primary visual cortex, physiological and behavioral reports seem contradictory. Neurophysiological studies on orientation tuning properties suggested that neuronal changes might come from decreased cortical local inhibition. However, behavioral results either showed no clear deficits in orientation processing in older adults, or proposed stronger surround suppression. Through psychophysical experiments and computational modeling, we resolved these discrepancies by suggesting that lateral inhibition increased in older adults while neuronal orientation tuning widths, related to local inhibition, stayed globally intact across age. We confirmed this later result by re-analyzing published neurophysiological data, which showed no systematic tuning width changes, but instead displayed a higher neuronal noise with aging. These results suggest a stronger lateral inhibition and mixed effects on local inhibition during aging, revealing a more complex picture of age-related effects in the central visual system than people previously thought.

Keywords: behavioral measure; computational modeling; early visual processing; neurophysiology; senescence.

Figure 1

Illustration of center-surround stimulus, orientation hypercolumns, population tuning curves and behavioral outcome. (A) Illustration of the stimulus, a center Gabor patch surrounded by an annulus of oriented grating, padded with orientation hypercolumns (red circles); the colored lines in right-top circle depict different preferred orientations of the local neuronal population; on the right, illustration of local interaction within an orientation hypercolumn, with three example neurons with different preferred orientations. Two inter-neurons, and their local connections (blue/green are inhibitory/excitatory connections); blue arrows depict inhibitory lateral interactions between hypercolumns. (B) Theoretical population orientation tuning curves (top) together with typical tilt repulsion curve (bottom) describing the orientation misperception of a vertical center stimulus as a function of the orientation of the surround; zero is vertical and positive values are clockwise tilts.

Figure 2

Examples of stimuli used and results of contrast sensitivity function and tilt illusion measures. (A) Schematic illustration of the contrast sensitivity function (CSF) measure; (B) Example of stimuli in tilt illusion measure. (C) Examples of results of CSF measures and fits in the 2-Alternative unforced Choice (2AuFC) design; squares depict the two chosen SFs for the subsequent tilt measures; (D) Averaged values of contrast sensitivities at the 11 different SFs measured for older adults and younger groups; (E) Tilt illusion results, indicated by orientation bias necessary to perceive the center as vertical, as a function of SOs and SFs (low-SF: circles; high-SF: triangles) for older adults (gray) and younger (black) groups; negative (positive) deviations for negative (positive) surround orientations correspond to repulsion; (F) Orientation thresholds around perceived verticality.

Figure 3

Relation between tilt repulsion, SF and CS at low-SF. Relations between tilt repulsion and SFs under ±15° (triangles) and ±30° (circles) SOs across all measured subjects in older adults (A) and younger (B) groups; Relations between tilt repulsion and CS at low-SF in the elderly group [±15° in (C); ±30° in (D)] and younger (±15° in (E); ±30° in (F)] groups. Error bars are bootstrapped SE.

Figure 4

V1 model illustration. (A–C) Example and prediction for orientation coding and decoding. (A) Uniform orientation tuning of the neuronal population. (B) Response of the neuronal population to center of 0°. and two different surround orientations of ±30° (top) and ±15° (bottom). (C) Orientation prediction of the model from the population responses for various surround orientations. Examples for three different set of parameters in red, black and blue. (D–G) Example and prediction for (SF) and contrast tuning coding and decoding. (D) SF tuning examples, with the characteristic tuning width decrease with increasing preferred SF. (E) Example of the relation between the minimum contrast semi-saturation constant and preferred SF. (F) Examples of CSF prediction for two sets of model parameters. (G) Examples of contrast response functions in the model for the minimum semi-saturation constant at few preferred SFs (arrows depict half-amplitude constant).

Figure 5

Model fitting results. Summary plots of subjects model parameters (I_inh, σ_θ, c_min) obtained from tilt repulsion data (A,B) and contrast sensitivity data (C) and relations to the SF (A,B) or peak contrast sensitivity (C,D). Insets depict histograms of the corresponding variable on the ordinate with significance values for the low-SF case.

Figure 6

Neurophysiological data re-analysis of orientation tuning. Examples of fitting results for orientation (A) and direction (B) of motion selective cells; OB, orientation bias index, DB, direction bias index. Distribution of minimum firing rate (C), amplitude (D), HWHA (E), and maximum firing rate (F) for the young and old cells. Arrows depict mean values.