Contrast normalisation masks naturalexpression-related differences and artificiallyenhances the perceived salience of fearexpressions.

Three datasets: Set 1: 'allphysicalblmdrms.xlsm' file. This is RMS contrast values measured for 140 KDEF face images of 5 facial expressions (neutral, fear, anger, happiness, disgust). Each expression has an RMS contrast value at each level of spatial frequency filtering (broad, low-, mid-, high-range spatial frequency). Set 2: 'amplitude_slope_data.xlsx' file. This is the averaged amplitude slopes for each facial expression, pooled across 140 KDEF faces. Set 3: Behavioural data from 19 observers. The amount of physical contrast required for each expression to perceptually match a reference target (whose Michelson contrast is set permanently at 10%). File 'RMS_matching-bsf-data.sav': apparent contrast values for 5 broadband expressions when upright and invertedFile 'RMS_matching-lsf-data.sav': apparent contrast values for 5 low-frequency filtered expressions when upright and invertedFile 'RMS_matching-msf-data.sav': apparent contrast values for 5 mid-range frequency filtered expressions when upright and invertedFile 'RMS_matching-hsf-data.sav': apparent contrast values for 5 ghih frequency filtered expressions when upright and invertedSet 4: 'skewness_all.xlsx' This is an excel file containing average skew values obtained from 16 KDDEF faces. These are provided for 4 frequency conditions (intact, low, mid, and high). Sheet 1 is the data for upright, natural images of faces (internal features only, grayscale). Data for their manipulated versions are available on sheet 2. This includes skew values for manipulated faces: those that were inverted by 180 deg and subjected to reversed luminance polarity. Faces were also internally cropped and grayscale.