This study investigates how perceptual thresholds for recognising faces and houses differ under varying levels of visual noise, using a General Linear Model (GLM) framework. We examine how individual electrode locations and their corresponding anatomical regions contribute to perceptual thresholds by analysing electrocorticography (ECoG) data. Channel-specific analyses reveal localised brain activity highly correlated with task performance, offering new perspectives on neural selectivity and perceptual robustness. Additionally, we compare GLM approach implementations using Python and MATLAB to evaluate differences in efficiency, accuracy, and analytical insights across computational platforms. This comparison highlights how analytical tools can influence the interpretation of neural data.