“…This approach allowed us to leverage the specific model predictions under the lateral inhibition, generalization, and linear hypotheses, and to quantify the fit between the empirical data and model predictions in one test across all conditions. The lateral inhibition pattern was expressed as the difference of two Gaussians (weights: +2,-2, +0.5, +1, +0.5,-2 for CS+, GS1, GS2, GS3, GS4, CS-), paralleling previous studies on visuocortical tuning (Antov et al, 2020;McTeague et al, 2015). The quadratic (weights: +2.5334, +1.0934,-0.0267, À0.8267,-1.3067, À1.4667) and linear (weights: +2.5, +1.5, +0.5,-0.5, À1.5,-2.5) trend were modeled after the analyses of the linear and quadratic component of the generalization gradient, which are commonly employed in the fear generalization literature (Ahrens et al, 2016;Lissek et al, 2014a;Lissek et al, 2014b).…”