Concerto for Intelligence Book Cover

Autistics Perceive Differently

In their research paper The Level and Nature of Autistic Intelligence II: What about Asperger Syndrome?, Soulières et al. (2011) venture the following assertion:

“Autistics [as compared to non-autistics] can maintain more veridical representations (e.g. representations closer to the actual information present in the environment) when performing high level, complex tasks.”

That statement falls within the Soulières et al. tradition of depicting nearly every aspect of autistic perception as an enhancement or advantage, while offering little to no consideration for the notion that some aspects of autistic perception might constitute a relative disadvantage—the latter omission an inexplicable form of blindness. Nonetheless, the above statement cannot be dismissed, for by targeting veridical representation as the measuring stick for evaluating perceptual differences, Soulières et al. are tapping into a long-standing literature showing that autistic individuals perceive their world not only differently than do non-autistic individuals, but that autistic individuals perceive their world in a manner that under an unbiased perspective might be described as more objective or more factual. Of course unbiased is itself a charged word, and its application here must bear directly on the accuracy of the Soulières et al. assertion. And so it is worthwhile to give the above statement some further consideration—both to highlight the ways in which it is informative, and to highlight the ways in which it is mostly missing the point.

To investigate the Soulières et al. assertion regarding veridical representation, it will be useful to lay out a hypothetical visual scene (something similar could be done in the auditory domain if so desired). Once this visual scene has been fully described, it will then be necessary to consider how three different types of entities might survey this scene, including one entity intended to represent autistic perception and another entity intended to represent non-autistic perception. But as will become apparent shortly, when it comes to experiencing the more veridical representation of the visual scene, it is going to be the third entity that emerges as the clear winner in capturing “the actual information present in the environment.”

Here is the scene: imagine a grassy field, similar to what one might come upon in a municipal park. Near the front of the field is a simple wooden bench where a woman and a young girl are seated and talking. Rising up behind them are four tall light poles, evenly spaced and situated so that they form a diagonal across the visual plane. The sky in the background is mostly blue with a few nondescript clouds straying past, and nothing else worthy of note is contained within the setting.

Here are three entities viewing this scene all from the exact same perspective, along with a rough description of what each perceives:

Although each of these descriptions is intended to highlight only general tendencies, most people would agree that the perceptions of Entity 1 closely match those of a camera, that the perceptions of Entity 2 are fairly typical of a non-autistic person, and that the perceptions of Entity 3 are more indicative of someone who might be on the autism spectrum. Each entity sees the exact same visual stimulus, but each extracts from that stimulus an entirely different set of information—the essence of what is meant by the concept perception.

By most ways of thinking, Entity 1 has by far the most veridical representation here—Entity 1 comes the nearest to perceiving this visual scene as it truly is. The key to a camera creating an accurate visual representation is, ironically enough, not to do much of anything at all with it, in particular not to impose any kind of structure or filter upon the visual material. No foregrounding of information. No backgrounding of unneeded data. No extracting of signal from sensory noise. Just reproduce the visual scene as it actually is—that is all that a camera is expected or required to do.

By contrast, both Entity 2 and Entity 3 come to their particular perceptions by imposing some kind of structure or filter on the raw visual stimulus they receive—which is to say that some of the features in their visual field will emerge as perceptual foreground, while other features will fade almost entirely unobserved into the sensory background. The perceptual process is quite similar for both Entity 2 and Entity 3 (and quite different from the perceptual process of Entity 1), but what distinguishes the perceptions of Entity 2 and Entity 3 is the emergent material of the signal itself. That is to say, there is a categorical difference in what tends to foreground in the perceptions of Entity 2 versus what tends to foreground in the perceptions of Entity 3.

Note that nothing has been said so far that would make it obvious that Entity 3’s representations are more veridical than those of Entity 2.

It is quite fortunate that Entity 1 is not a biological or responsive agent. If it were such an agent, having the most veridical representation of a visual scene would actually manifest as a huge liability. To be responsive to an environmental stimulus requires that information be extracted from it, exactly what Entity 1 is not designed to do. This has actually been a problem in the domain of robotics, where despite the availability of extremely accurate cameras, it has been nonetheless difficult to get machines to respond flexibly and constructively to various visual stimuli, precisely because it has been difficult to get machines to recognize what constitutes the essential foreground of a visual scene and what needs to be dismissed as inconsequential background. Possessing a perfectly veridical representation of an environmental stimulus is tantamount to experiencing sensory chaos; everything comes across as noise, nothing emerges as signal. And with no signal, there is no useful information, and with no useful information, there is no ability to respond with purpose.

This observation regarding the liability of a perfectly veridical representation is relevant to more than just the perceptual characteristics of Entity 1 (which after all, not being a biological agent, is not really vulnerable to the liability). As will be seen shortly, it is the threat of this liability that plays a critical role in the development of Entity 3’s perceptual characteristics, characteristics that are motivated almost entirely by the need to avoid the debilitating sensory chaos that arises out of a perfectly veridical representation.

First, however, it is necessary to take a detour and consider further the perceptual characteristics of Entity 2 (non-autistic perception). By even the most casual observation, it is clear that the primary characteristic that delineates and defines non-autistic perception is its strong bias towards foregrounding human features within a sensory environment. Human faces, human voices, human movements, human actions. This is an instance of conspecific awareness and recognition, a trait common throughout the entire animal kingdom, whereby each organism is perceptually drawn primarily to the other members of its own species. In Homo sapiens, it is this human-specific perceptual focus that provides much of the necessary structuring and filtering that allows a person to extract meaningful and actionable signal from what would otherwise be sensory noise. The effects of this human-specific focus can be witnessed in the depiction of Entity 2’s perception of the given visual scene, where attention is drawn first and foremost to the people within the scene, and in an impressively detailed manner. Even those features of the scene not directly connected to humankind are still brought to the foreground by an association to the species (the light poles for instance are comprehended as helping people see at night and not be afraid).

People-centric representations form much of the common thread that draws non-autistic cognition into a cohesive whole, and this common thread works on more than just an individual level. Since nearly every human shares this strong bias towards people-centric representations, these representations serve to coalesce not just individual thoughts and behaviors, but also the conventions and activities of the population as a whole. As is the case with nearly every other species upon this planet, the cohesion brought about by conspecific foregrounding delivers to humanity an entire array of biological and evolutionary benefits, directly supporting many survival and procreative efforts. And thus in a very real and observable way, non-autistic perception bestows to its possessors an essential and useful advantage, an advantage it would be folly to ignore.

The perceptions of Entity 3 (autistic perception) differ from those of its non-autistic counterpart precisely in the comparative lack of a human-specific focus. The difference is not absolute—autistic individuals to varying degree carry some level of natural attention for other people—but the difference is always large enough to produce significant effect, implying conspecific distance. The natural structuring and filtering of sensory experience provided by human faces, human voices, etc. is not strongly acquired in autistic individuals; for them, the coalescing impact of conspecific foregrounding goes mostly absent. As a result, autistic individuals find themselves in nearly the same circumstance as that of Entity 1, as that of a camera. Sensory impressions are received more raw and unfiltered, the details of the sensory field are processed more indiscriminately, and thus just as Soulières et al. would suggest, autistic perception approaches much nearer to that of veridical representation.

In this form, however, veridical representation does not offer any advantage or enhancement for autistic individuals; instead it manifests as a dangerous liability. Autistic individuals, quite unlike cameras, are biological agents, and to remain viable as biological agents they must be able to respond to their sensory environment with constructive purpose, they must be able to extract sensory signal and to ward off sensory noise. Many of the developmental and sensory challenges that autistic individuals must endure and overcome can be traced to the inherent sensory chaos that accompanies a perfectly veridical representation, and to the extent that autistic individuals do achieve a level of advantage and enhancement from their condition, it is not because they come nearer to veridical representation but rather because they manage to escape it. Closed off from the coalescing benefit of conspecific perceptual foregrounding, autistic individuals must unify and organize their sensory experience by foregrounding instead an alternative target, and the target that autistic individuals almost invariably gravitate towards is the structural framework of the non-biological world. Pattern, number, repetition, symmetry: the same elements that break the background chaos of the non-biological world are the same elements appropriated by autistic individuals to help break the sensory chaos of their perceptual experience. This propensity to foreground the structural features of the surrounding environment is evident in the depiction of Entity 3’s perception of the given visual scene, where attention is drawn first and foremost to the non-human artifacts, with an emphasis on the characteristics of pattern, symmetry, structure and form (the spacing and arrangement of the light poles for instance). Furthermore, in recalling how Entity 2 grasps the non-biological features of the visual scene by associating those features back to humanity, one can observe how Entity 3 effectively reverses this process, comprehending the people within the visual scene not by conspecific affinity, but instead by categorizing such people under structural characteristics, such as number and relative size.

In their pure form, autistic perception and non-autistic perception represent two different aspects of human perception that have now become so blended within the human population that in reality a pure version of either form is rarely encountered. Non-autistic perception stems from humanity’s animal past and carries with it the strong and ancient influence of conspecific awareness and recognition, providing much of the binding that holds the species together and much of the vigor that drives the species to greater conquest. Autistic perception represents the atypical characteristics of humanity’s recent perceptual turn, an unprecedented opening onto the world’s structural and non-biological features, providing a means for overcoming many of the contingencies of survival and procreation, and providing the constructive vision for an exploration of a massively sized and massively detailed universe. Each form of perception offers an advantage the other cannot provide, and the strength of overall human perception has been its ability to combine the benefits of these two discordant constituents. For the species as a whole, the result of the combination of these two forms of perception has been astounding and prodigious, for quite unlike any other species—and quite unlike humans themselves from not that long ago—modern humans have been able to break the chains of biological constraint and are now rapidly building for themselves an environment of essentially unlimited potential.

To the extent that autistic perception flirts with the dangers of veridical representation, and to the extent that autistic perception ultimately draws its organization and unification from the structural features of the non-biological world, it is perhaps not unreasonable to suggest that autistic perception has something about it that can be described as more objective or more factual. But it is just as important to recognize that in its final form, autistic perception stands in defiance of veridical representation, and that in its structuring and filtering of environmental stimulus, autistic perception remains entirely analogous to its non-autistic counterpart. In truth, there is no unbiased perspective from which to judge the accuracy or advantage of differing forms of perception: each form of perception will be predisposed to favoring its own particular way of perceiving things. But when it comes to human perception, what matters is not the relative advantage or disadvantage of its constituent parts, what matters is the power of their amalgamation—the key to humanity having broken free of its former biological limitations, and the key to the world’s structural framework having found a means to come to life.

Thus the lesson to be learned from autistic perception is simply this: autism is a perceptual difference, induced by the threat of a perceptual liability, resulting finally in a perceptual enhancement for the species as a whole.

Concerto for Intelligence home page

Copyright © 2017 by Alan Griswold
All rights reserved.