What is a perceptual stimulus? Many philosophers never ask themselves this question. They often take it for granted that vision science has settled the issue and nothing more needs to be said except that
- Visual perception depends on light rays hitting the back of the retina, forming an “inverted image”.
- The subsequent “retinal image” is imperfect and ambiguous in respect to “macro properties” since it is more or less “flat” , “upside down”, and registered by messy biological tissue.
- The brain has to “compensate” for the ambiguity by making “hypotheses” about the macroworld. The first stage of processing is the “2D primal sketch”. Next, the brain generates 3D “depth” from the more immediate primal sketch. Thus, the brain is essentially a Helmholtz “inference machine” from the ground up.
- Perceptual experience is what-its-like for the brain to make inferences. Visual experience is thus a simulation grounded by the immediate retinal registration.
- Our intentionality is directed towards the simulated inferences rather than the world itself. Visual experience is thus illusionary and generated “internally” by the brain, hence the name internalism.
I take it that this crude picture is more or less an accurate representation of internalism. These presuppositions have become so entrenched, they hardly come up for review. I think it is high time to examine them, particularly in respect to the taken for granted assumptions about what constitutes a perceptual stimulus. The stimulus is often understood as being the immediate rays of light “bumping into” the retina. The retina is said to register the light strictly in terms of the wavelength and intensity i.e. the “primary qualities” of the physical sciences. Thus the retinal image corresponds to “micro” properties rather than molar properties. And because the registration of the retinal image is in terms of micro properties, molar properties like depth, surface, texture, etc. need to be “inferred” and are not directly perceived by the brain. It is said then that the micro properties are what make perception “ambiguous” and prone to error in respect to the macroworld.
But why should we assume that the brain is directed towards micro properties? It seems perfectly reasonable to suppose that the brain is behaviorally sensitive to molar properties specified in the ambient array of light “filling” normal environments. After all, molar properties are where all the action is. Only a programmed computer would be interested in micro properties. Biological bodies are far more interested in what’s going on at the molar level, where temporal events occur. Indeed, as J.J. Gibson points out, “Animals and men [directly] perceive motions, events, episodes, and whole sequences”.
And because all the episodes relevant to survival happen at the molar level (being attacked, hunting, socializing), it’s reasonable to suppose that the brain would have evolved so as to be directly sensitive to molar properties rather than micro properties. And if this is the case, then we need not assume that the brain is a Helmholtzian inference machine at the ground level of visual perception. Contemporary ecological optics has repeatedly demonstrated that molor properties such as depth, surface, and texture are directly specified by the invarient patterns of stimuli resulting from movement through the ambient optic array. The transformations across the retina specify molar properties in virtue of successive and adjacent order in the ambient optic array. Perception is thus capable of “holistic” sensitivity from the bottom up. Natural selection insured that brains developed the ability to be directly attuned to molar properties like motion, not micro properties like wavelengths and intensity. Motion indicates prey or predator. Wavelength indicates nothing unless you are a scientist.
We can thus overturn the five assumptions stated above. Intentionality is directed towards molar properties in the world rather than representations in the head. Because internalists are empirically mistaken about the ambiguity of perceptual stimuli available in the optic array, the absolute bottom-up inference making of classic computational approaches can be rejected as overcompensatory. Sure, the brain probably does make many inferences, particularly at higher-level stages in visual processing. But a complete Helmholtz machine? Unlikely. The availability of molar information in the optic array “ripe for sampling” obviates the need for a bottom-up inference from the micro to the molar. Our genetics ensure that nervous tissue is behaviorally sensitive to molar properties in the environment. This means that the “eternal now” of instantaneous registration is mythical and purely heuristic. Perceptual registration is temporally extended in virtue of its intentional directedness towards molar properties. We need to thus think about the brain in terms of a temporally extended coherence or “resonance” that is behaviorally sensitive to molar properties. This requires rethinking the constitution of perceptual stimuli and the evolutionary development of sensorimotor systems.