Keywords

detection, contrast energy, spatio-temporal models

Abstract

The Barten (1994) spatial-temporal model was used to predict the Gabor stimulus contrast energy thresholds reported by Carney et al. (2013). The RMS error of fit was 1.6 dB, corrected for the number of parameters (6) estimated. The model has two lowpass spatial-temporal channels, combined by inhibition as in our spatial models (Watson & Ahumada, 2005; Ahumada & Watson, 2013). Computation of models predictions were greatly simplified by the spatial-temporal separability of the stimuli and the simplifications that result from using Gaussian filters in the spatial domain. The best fitting spatial filter frequency cutoffs are 11.4 and 0.88 cpd. The temporal filters are Gamma with time constants both set to 12.5 msec, with the number of exponential components equal to 3 and 2. The ideal observer noise spectral density estimate is 4.3 dBB (dB re 10^-6 deg deg sec).

Start Date

17-5-2017 10:32 AM

End Date

17-5-2017 10:54 AM

Share

COinS
 
May 17th, 10:32 AM May 17th, 10:54 AM

Spatial-Temporal Visible Contrast Energy Predictions of Detection Thresholds

The Barten (1994) spatial-temporal model was used to predict the Gabor stimulus contrast energy thresholds reported by Carney et al. (2013). The RMS error of fit was 1.6 dB, corrected for the number of parameters (6) estimated. The model has two lowpass spatial-temporal channels, combined by inhibition as in our spatial models (Watson & Ahumada, 2005; Ahumada & Watson, 2013). Computation of models predictions were greatly simplified by the spatial-temporal separability of the stimuli and the simplifications that result from using Gaussian filters in the spatial domain. The best fitting spatial filter frequency cutoffs are 11.4 and 0.88 cpd. The temporal filters are Gamma with time constants both set to 12.5 msec, with the number of exponential components equal to 3 and 2. The ideal observer noise spectral density estimate is 4.3 dBB (dB re 10^-6 deg deg sec).