To the Editor: Dr. Loddo’s speculation is that the critical variable accounting for variable results in the visual scanning of faces by individuals with autism is "level of difficulty." Paradigms using static faces as stimuli are associated with more normative patterns of visual scanning because they place a lower level of stress on the "capacity of the mental apparatus." In contrast, paradigms using dynamic (i.e., video) stimuli place added burden, and therefore, individuals with autism move their attention to "segmentary scanning of lower and/or peripheral parts of the face…in order to keep inputs at a processable level." His hypothesis is that these paradigms stay in a continuum of difficulty, and he theorized some values for the various paradigms as well. His hypothesis is not one of overwhelming affective burden (e.g., too much arousal) but one of overwhelming cognitive burden. Although this is a possibility, we, and apparently other research groups, have no evidence to support or contradict this hypothesis. It is an empirical question that would require some measurement of "burden on mental processes." What we can state, however, is that there is evidence to suggest that typically developing children as young as 4 days old prefer to look at eyes that are looking at them and that by 3 months of age, they are preferentially looking at people’s eyes (rather than mouths). Also, in nature, babies do not look at static faces but at dynamic faces moving and talking to them. Therefore, there is some evidence that the level of burden on mental capacities that viewing complex social stimuli exposes people to cannot be too overwhelming. More important, however, we have at present no evidence that looking at static faces in behavioral and neuroimaging experiments and looking at naturalistic social approaches (or our research group’s attempt to simulate such situations in our eye-tracking laboratory), in fact, correspond to the same behavioral/cognitive demands and can be mapped onto the same neural systems. This is why our research effort has been to allow for great complexity in the stimuli used in our paradigms while developing highly sophisticated ways of measuring viewers’ eye-scanning responses to those stimuli. Typically, researchers have reduced paradigm complexity in order to avoid confounds because measurements were relatively crude. The eye-tracking technology and the methods that we have developed to analyze eye-tracking data give us an opportunity to change this.