From coarse to fine? Spatial and temporal dynamics of cortical face processing

Abstract

Primary vision segregates information along 2 main dimensions: orientation and spatial frequency (SF). An important question is how this primary visual information is integrated to support high-level representations. It is generally assumed that the information carried by different SF is combined following a coarse-to-fine sequence. We directly addressed this assumption by investigating how the network of face-preferring cortical regions processes distinct SF over time. Face stimuli were flashed during 75, 150, or 300 ms and masked. They were filtered to preserve low SF (LSF), middle SF (MSF), or high SF (HSF). Most face-preferring regions robustly responded to coarse LSF, face information in early stages of visual processing (i.e., until 75 ms of exposure duration). LSF processing decayed as a function of exposure duration (mostly until 150 ms). In contrast, the processing of fine HSF, face information became more robust over time in the bilateral fusiform face regions and in the right occipital face area. The present evidence suggests the coarse-to-fine strategy as a plausible modus operandi in high-level visual cortex.

Figure 1.

(a) LSF, MSF, and HSF faces were presented at 3 exposure durations, immediately followed by a Gaussian mask. The phase of face stimuli was either intact or scrambled in the Fourier domain. All conditions were equated for luminance, RMS contrast, and spectral composition. They were randomly interleaved within a run and subjects categorized each trial as an intact or a scrambled one. (b) Performance accuracy in intact-scrambled categorization was at ceiling and was not influenced by SF, exposure, or stimulus factors. In contrast, correct response times were shorter for intact than scrambled conditions, and significantly increased at 300-ms exposure duration compared with 75- and 150-ms exposure conditions.

Figure 2.

Amplitude spectrum as a function of SF in unfiltered, LSF, MSF, and HSF stimuli, before and after luminance and RMS contrast have been equalized. Note that luminance and contrast equalization did not alter spectral envelope.

Figure 3.

Left: Fourier amplitude is plotted as a function of orientation, revealing the similar orientation content across intact and scrambled conditions in each SF conditions, separately. These plots are based on a single measurement, so not taking into account the lack of a set of continuous orientation vectors in the Fourier domain (e.g., Hansen and Essock 2004). Right: Fourier amplitude plotted as a function of SF. Note the high similarity between intact and scrambled spectra.

Figure 4.

Average BOLD activity in the rFFA. (a) Normalized beta weights in the rFFA (bars 5 mean intrasubject variance). (b) Effect size of the difference between intact and scrambled faces in separate SF and exposure duration conditions. (c) Grand averaged event-related time course of intact and scrambled face processing in the rFFA. Time courses are expressed in percent signal change relative to fixation baseline activity (baseline interval: from 2 to ±2 TR around preparatory cue onset). The activity time courses shown on (c) reflect the findings based on the beta weights.

Figure 5.

Effect size plots in (a) face-preferring regions (lFFA, rOFA, rSTS, lSTS, bilateral AIT) and (b) object-preferring regions (right and left ventral LOC).