1 Introduction
Science seeks, among other things, unity in diversity. One goal of the theoretical scientist is to find unifying structures and causal laws which encompass, as special cases, the explanations accepted for specific phenomena or or properties of individual systems. Behind (e.g.) the diversity of atomic and subatomic phenomena, from the gravitational attraction of atoms to the chromatic properties of quarks, theoretical physicists seek a unity, a unified field theory, which encompasses as special cases the explanations accepted for these phenomena. Similarly, behind the diversity of possible algorithms, from the recognition of primes to the scheduling of traveling salesmen, computer scientists have found a unity of structure, the Turing machine, which encompasses as special cases all algorithms.
But behind the diversity of perceptual capacities (e.g., stereovision, auditory localization, sonar echolocation, haptic recognition) no such unity has been found. The field of perception has no unifying formalism remotely approaching the scope and precision of those found in physics and other natural sciences. This is perhaps not surprising. Before one can unify one first needs something to unify. In the case of perception one first needs theories of specific perceptual capacities that (1) are mathematically rigorous, (2) agree with the empirical (e.g., psychophysical) data, and (3) work. And these have, until recently, been in very short supply.
But there is now reason for guarded optimism. The last few years have witnessed the genesis of just such theories. We now have theories of (e.g.) stereovision that are mathematically rigorous, that are not too comical to the psychophysicists, and that actually (sometimes) work when one implements them in computer vision systems. Theories with similar salutary properties are on offer for aspects of visual-motion perception, the perception of shading and texture, object recognition, and light source detection. With this growing collection of rigorous theories comes a growing temptation: viz., the temptation to wade around in this collection of theories in search of structural commitments that are common to them all. If such we find, from these we might fashion a unifying formalism which encompasses each theory, perhaps every perceptual theory, as a special case.
We have succumbed to the temptation. And, as you might have guessed, we think we have found something. This book records where we have looked, what structural commitments we have encountered in theory after theory, and what unifying structures we have, in consequence, constructed.
Perhaps the most fundamental is a structure we call an “observer.”1 An observer is, roughly, the static structure common to all theories of perceptual capacities we have so far studied. Much of this chapter and the next are devoted to the explication of this structure, so we shall not dwell on it here. Instead we shall enter claims and disclaimers regarding this structure.
First a disclaimer. There are, of course, many perceptual capacities whose theories we have not yet studied, and far more capacities, e.g., in the modalities of taste and smell, for which there simply are no adequate theories. Our own training is in visual perception, with the consequence that the examples adduced throughout this book are primarily visual.
Now for a claim. To make things more interesting, we shall stick out our necks and advance the definition of observer as a unifying structure not simply for some capacities in vision but, rather, for all capacities in all modalities. Accordingly, we propose the following observer thesis: To every perceptual capacity in every modality, whether that capacity be biologically instantiated or not, there is naturally associated a formal description which is an instance of the definition of observer.
This thesis is vulnerable to disconfirmation by counterexample. As new capacities are studied, or as the structure of existing theories of specific capacities are reexamined, capacities may be found whose formal structures are not instances of the definition of an observer. And given the definition’s foundation in a somewhat small collection of specific theories this eventuality is, despite our efforts to the contrary, not impossible. If it happens, then the definition will be, in consequence, further refined or entirely replaced by a more adequate structure.
After defining an observer in chapter two, we set it to work on several problems in perception and cognitive science. One problem is to define the concept transduction. Some relevant intuitions here are that transduction involves the conversion of energy from one physical form (say light) to another (say neural impulses); that transduced properties are, in a certain sense, illusion free; that in the case of vision it is properties of light that are transduced and the transducer is the retina; and that in the case of audition it is properties of sound that are transduced and the transducer is the cochlea. But turning such intuitions into a workable definition has proved difficult; it is a remarkable fact about the field of perception that such a basic concept is as yet ill-defined. It indicates, perhaps, that not all the relevant intuitions can simultaneously be granted. Indeed, some get sacrificed in the observer-based definition we propose.
We also employ observers in an effort to define the theory neutrality of observation. Philosophers still debate about the proper intuitions for this term: some argue that to say observation is theory neutral is to say that the truth of observation reports is independent of any empirical hypotheses; others argue that it means that scientific beliefs do not “cognitively penetrate” perception, i.e., roughly, that the beliefs one holds do not alter one’s perceptual apparatus—the intuition here being that if observation is in this sense theory neutral then two scientists could hold competing theories and yet agree on the data that they observe in critical experiments. We employ observers not to settle the empirical issue (viz., is observation in fact theory neutral) but, rather, simply to define it. To this end we first propose relational definitions for the terms cognitive and cognitive penetration. We then formulate the claim that observation is theory neutral to be the claim that the relation cognitive is, in the appropriate context, an irreflexive partial order. This development, together with the definition of transduction mentioned above, leads to a novel functional taxonomy of the mind. This taxonomy is discussed briefly in chapter two and more extensively in chapter nine.
Observers capture, so we claim, the static structure common to all perceptual capacities. But perception is notably active: it involves learning, updating perspective, and interacting with the observed. To account fo...