Abstract
The tilt aftereffect (TAE) is taken to reflect short-term experience effects on perception, on the scale of seconds to minutes, but the rules governing the temporal integration of the underlying causes are not clear. TAE is explained by local effects attributed to accumulated sensory adaptation, but higher-level effects depending on statistical regularities in the adapting sequence are also implicated (Pinchuk-Yacobi, Dekel, & Sagi, 2016). Here we attempt to isolate short-term and long-term effects in the generation of TAE. We used a massed repetition design (Adaptor-Test trials: "AT", adaptors oriented ±20°, randomly mixed, N=6 observers) to measure the magnitude and spatial selectivity of TAE for briefly presented peripheral Gabor patches (Adaptor and Test duration=50ms, ISI=600ms, λ=0.3°, σ=0.6°, contrast=50%, eccentricity=2.1°). Results showed standard retinotopic TAE ("ipsi": 1.2±0.55°, Mean±SD), non-local TAE (when adaptation and test at opposite sides of fixation, "contra": 0.6±0.5°), and a surprising correlation between ipsi and contra (R2=0.78, p=.02) permitting a stable description of TAE in terms of its spatial selectivity (ipsi-contra, M=0.6±0.25°, Mean±SD). Next, we used a sterilized repetition design ("V--AT", N=4 observers), where a vertical Gabor presentation ("V") and a pause ("--", duration=1200ms) precedes each AT event. This design showed a larger local TAE (1.7±0.7°, Mean±SD), but a somewhat reduced contra TAE (0.3±0.4°), showing a stronger spatial selectivity (ipsi-contra: 1.4±0.46°, Mean±SD, p=0.03). Importantly, AT and V--AT measured identical TAE when mixed in blocks ("V--AT/AT/AAAAT" design, N=4 observers), equating long-term history. Thus the measured AT vs V--AT differences are due to a mechanism operating on slow timescale, possibly reflecting long-term statistics. We conclude that TAE has a short-term local component, probably sensory, and a longer term component reflecting stimulus statistics over longer time.
Meeting abstract presented at VSS 2018