The debate between direct perception and indirect perception has been going on for quite some time. Indirect theorists often point out that anatomical facts such as afferent and efferent nerves undoubtedly indicate that perception is indirect because simple anatomy tells us that the stimulus has to first be transduced and shuttled through the various nervous channels before being cognitively processed and transformed into a genuine “perception”. But if we are going to make theoretical progress , we must realize that anatomical facts will not settle this debate.
Direct theorists have never denied that the perceptual process can be artificially decomposed into anatomical facts. Both sides can agree that a stimulus must pass through various nervous channels; it does not get a free pass straight to the mind, the stimulus must be mediated by the brain. But what is the nature of this stimulus? Theorists on both sides rarely make their definition of stimulus explicit. It is assumed that everyone knows what everyone else means when they talk about the perceptual stimulus. This is a mistake. The issue is much more complicated that it first seems.
Indirect theorists often start their psychologizing from the perspective of neuroanatomy and physiology. They first zoom in on the retina very close and attempt to build a psychological model of vision beginning with meaningless physical intensities as proposed in the physical sciences. It is usually assumed that any psychological theory of visual perception must explain how the brain interprets this raw physical data (“sense-data”) and converts it into a meaningful percept. Sometimes this transition from meaninglessness to meaning is talked about in terms of the generation of true beliefs or true representations. But the essential question is always, How do you go from raw physical data to meaningful perception when the meaningless physical intensities are highly ambiguous and often irrelevant?
Direct theorists also make a distinction between meaningless sensation and meaningful perception, but reject the idea that the perceptual stimulus is meaningless. The classic example is the Ganzfeld experiments. 20th century vision scientists discovered that if the physical stimulus is undifferentiated, meaningful perception fails to occur even though the visual system is being stimulated. Direct theorists thus make a distinction between sensory stimulation and stimulus information. Imagine standing in an open field on a bright, cloudless day. When you orient yourself such that the sky fills your entire visual field, your sensory receptors are being stimulated but there is no meaningful perception occurring because there is no meaningful information to be differentiated from the stimulus because the stimulus is entirely homogeneous and undifferentiated. In other words, the undifferentiated sky contains no stimulus information, although it is stimulating.
Now here is the important point. The facts associated with sensory stimulation are facts of an anatomical or physiological nature. But they are not psychological facts. We cannot decide between an indirect theory and a direct theory on the basis of these physiological facts. We must focus on the perception of meaningful stimulus information.
Indirect theorists explain meaningful stimulus information with a mix of association psychology and computational representationalism. Meaningful percepts are generated whenever the cognitive system makes certain inferences (associations) from the raw stimulus with the premises either innate or learned in experience. Classic cognitive science talks about explicit symbol systems and generalized intelligence, but modern computational stories have become more and more complex. But almost all of them assume that the quintessential problem for visual perception is turning meaningless data into meaningful perceptions. This is nothing less than the mind/body problem applied to visual science.
But direct theorists reject this approach altogether. Although direct theorists admit that stimulation is sometimes meaningless (such as when we are looking at the undifferentiated sky or in a snow storm), they emphatically insist that, under normal circumstances, the immediate terrestrial environment is differentiated and highly organized. The differential structure of the ambient energy fields surrounding an organism is informationally rich. But not in the Shannon cybernetics sense of information (which was never meant to be a psychological theory). The environment is informationally rich insofar as it contains information specific to affordances.
Direct theorists claim that the ambient energy fields are filled to the brim with redundant information specific to affordance properties. Affordance properties are real, objective facts about the environment. I thus disagree with Chemero and side with Reed on the ontological status of affordances. On my view, affordances are real properties of the environment that persist through time. The fact that the ground will afford my locomotion upon it is a fact that is independent of whether I actually utilize the ground for the purpose of locomoting. But it would be a mistake to think that this fact about the world is a molecular or local fact. The fact that the ground surface supports locomotion is a molar fact.
If we look at the ground on the timescale of millions of years, the ground is but a ceaseless flow of energy, ever shifting and changing. On the ecological timescale, however, the ground is stratified, ossified, and stabilized. And since our perceptual systems are tuned into this ecological scale, we do not perceive the molecular flux of the ambient energy fields. The ground is perceived as a continuous rigid surface with the property of “supportability”. This is an affordance-property. The detection of such properties by the nervous system is highly useful. We can expect that evolved systems would be optimally tuned to detect these properties because they are facts about the environment most relevant to survival.
We can cash this out psychologically in terms of how the perceptual systems seek information specific to these affordance properties. The affordance-property of supportability is a persisting fact about the ground surface. Under normal evolutionary conditions, the perception of this affordance-property is done so as to coordinate the motor system and enable successful navigation through the terrestrial environment.
Take Herbert Simon’s example of an ant crawling along the beach surface. On first blush, its locomotive pathway seems highly complex and difficult to explain.The indirect theorist would attempt to explain its locomotive patterns in terms of internal control wherein the motor sysem is totally in charge of directing where to place each leg. However, the direct theorist would explain its locomotive patterns by saying that the ant is merely following the contours of the sand. Rather than the ant controlling itself from within, the environment is guiding the ant. Put another way, the ant is using the affordance-properties of the beach to coordinate and regulate its behavior. The pathway looks complex only because the sand surface is complex, but the psychological control is actually quite simple.
On the neural level, we can say that there is a intrinsic flexibility and variability in the nervous system, otherwise the system would never be able to handle the complexity and novelty of the ever changing environment. However, the persisting affordance properties of the environment are sought out and detected so as to help coordinate motor behavior. Rather than the perceptual stimulus being a raw mechanical instruction, the perceptual stimulus helps “select” or “trigger” useful patterns of neural activity from the intrinsic variability. Faced with the same tasks and problems over a developmental life cycle, certain patterns are going to be burnt in that help the animal cope with the environment. But it would be a mistake to decompose the task of action-coordination into purely internal neural circuity. The affordance theory recognizes how animals use both internal and external means of coordinating behavior. The neural system readily uses information specific to affordances to regulate behavior. This means that some behavior control is “external”. The problem then is not, How does the brain generate meaning from meaningless data? Rather, the problem is, How does the brain seek out meaningful information and then use it to regulate and coordinate its autonomous behavior?