The invention pertains to a Model based detection of attention distribution and/or reaction times and further effects of human-machine interfaces, comprising for a plurality of test-persons the steps of• Determining and labeling (S100) areas of interest (AOI) and their associated parameters (P), which are selected from the group comprising the expectancy of information appearing within the AOI, the value of all tasks for which the information appearing within the AOI might be relevant, and the relevance of the information appearing within the AOI for each of the aforementioned tasks, per test-person of a human-machine-interface to be evaluated,• Clustering (S200) the determined areas of interest (AOI) and their associated parameters (P: information expectancy, task value, relevance) for a plurality of test-persons, based on the determined area and/or their labels and/or their associated parameters (P) into classes of area of interest (AOI-classes),• Simulating (S300) gaze behavior and reaction times per test person based on areas of interest (AOI) and associated parameters (P) defined by the respective person,• Grouping of simulated eye fixations and saccades for a plurality of test persons according to the determined classes of area of interest (AOI-classes), whereby based on said data for a plurality of test persons probabilistic predictions relating to the distribution of visual attention and/or gaze transitions and/or reaction times are generated,• Visualizing predictions and determining the causes for predicted distribution of visual attention and reaction times, by analyzing and visualizing associated parameters (P: information, expectancy, task value, relevance) of areas of interest (AOI) for a plurality of test persons,• Identifying (S600) inconsistent input data, concerning definitions of areas of interest (AOI) and definitions of associated parameters (P) for multiple test persons to identify potentially different understandings of the evaluated system acr
