To the Editor: I wish to add a few remarks to the interesting discussion between Ami Klin, Ph.D., et al. +(1, +2) and Chantal Kemner, Ph.D., and Herman van Engeland, Ph.D., M.D. +(3), focusing on the levels of difficulty that social visual fixation poses to persons with autism. Considering a scale from 1 to 10, I will arbitrarily assign level 8 to the paradigm of Dr. Klin et al., the observation of filmed social scenes. Proportionally, an individual could experience level 9 while being engaged in a real, familiar social situation and level 10 in an unfamiliar one. As for the experiment of van der Geest et al. +(4), level 3 could be attributed to looking at drawings in which human figures are not so rich in emotional and expressive data (p. 72 of their article).

At an intermediate level are tasks in which the probands’ visual scan paths of photographs of expressive human faces are recorded +(5). The graded results registered at these three different levels suggest that the richer the data to be processed, the less that relevant parts of scenes and faces are attended by viewers with autism. In the study by van der Geest et al. +(4), low levels of difficulty permit autistic probands to look at human faces and objects in a way that is comparable to that of normal subjects. In the paradigm of Dr. Klin et al. +(1), on the other hand, the presence of fluid, emotionally laden visual and verbal interpersonal exchanges overburdens the capacity of the mental apparatus. Individuals with autism automatically move their attention from the zones with the highest content of holistically explorable data (the eyes-nose-mouth area) to the segmentary scanning of lower and/or peripheral parts of the face, seemingly in order to keep inputs at a processable level. They rarely shift from one character to another for the same reason.

I think that in autism a central system of integration of brain functions is at fault, and the consequent reduced web of activation excludes higher abilities, such as the recognition of (complex) social cues. As a consequence, the functional development of cortical areas dedicated to that kind of global simultaneous processing (e.g., the fusiform facial area) is never facilitated because fully using these areas would absorb an excessive amount of mental energies at the expense of more basic abilities, and confusion would ensue. Areas dedicated to a more parsimonious, piecemeal computation of inputs (e.g., the inferior temporal gyrus) are alternatively employed.