The Cartesian Broadway
The `Cartesian theatre' is a term for the arena of consciousness---the active processing in which concepts interact and thoughts are juggled. A particular species of process in this theatre is that of filling-in, the process by which we perceive continuity of the world across regions with no neural response, such as the blind spot of each eye. Dennett (1991) has argued that there is no need for filling-in because the absence of stimulus input is not a positive indication of a lack of input.
Pessoa, Thompson, and Noe (1998) maintain that Dennett's argument
for the lack of need for filling-in misses important aspects of
perceptual processing but that the mechanisms for filling-in proposed
by Grossberg (1983), for example, are unnecessarily elaborate. They conclude
that the `personal' level of analysis, of the observer interacting with
the environment, is the appropriate viewpoint for understanding perception
in general and filling-in in particular. Although the article makes
excellent points against the need for a mechanism of filling-in,
it throws out the baby of neural specificity with the bathwater of isomorphism and the homuncular observer. It seems incontrovertible that the core
act of perception is sensory processing by a stationary observer. The complexity of intracortical interconnectivity still allows
local specificity in the representation of higher-order stimulus
properties.
Sparse sampling obviates the need for filling-in
In principle, perhaps, Dennett is right.
There is no need for filling-in of stimulus regions
where no seeing occurs. The classic example is in cases of sparse sampling.
The sky looks a uniform blue, even though the `blue'
photoreceptors are spaced far enough apart that the gaps
could easily be seen by the intervening bipolar cells fed by the `red'
and `green' photoreceptors. So we have sufficient optical and neural resolution to see the sky as a grid of blue dots; why do we see it as a uniform
blue? Dennett's answer would be that we cannot know that there is no blue
in the spaces between the receptors if the subsequent neural processing
samples only the output of those receptors. We do not need to fill-in
the gaps because we do not survey the gaps.
Although this is a logically acceptable conclusion, it is by no
means self-evident in the case of blue, because the `red' and `green'
receptors are required to generate the blue percept
by means of the blue/yellow opponent pathway. Presumably, there is coarser spatial integration in this pathway to limit its overall sampling
to the level provided by the `blue' receptors. But this now becomes an
empirical rather than a philosophical question.
The poverty of linking propositions
Incidentally, this example plays havoc with
Mach's principle that the physiology accounts for the perceptual state, Kohler's isomorphism, and Teller's linking propositions. Taken
literally, the perceptual state appears continuous, the physiological
sampling is discrete; therefore the cells cannot account for the percept!
It is only when considered in the context of subsequent processing that
the link makes sense.
The converse is also true: a simplistic interpretation of linking propositions would take psychophysiological parallelism as a sufficient condition for probable identity. Just because a physiological response decays at the same rate as a percept, however, one would not rush to assume that one explains the other. A contemporary example is provided by functional magnetic resonance imaging (fMRI) of the brain, which records properties of blood flow. The parallelism between blood flow dynamics and, say, perceptual aftereffects may be extensive, but no-one believes that the percept is carried by the blood. It seems that a larger set of linking propositions is required; for example, one has to have a system capable of information transfer and storage before it is plausible that it could support conscious percepts. Thus, one would look for such a system as the causal agent of the blood flow effects, as is available in the form of the network of neurons in the brain, rather than attributing perceptual effects directly to the measured variable.
Thus, it is easy to agree with Pessoa, Thompson, and Noe that psychophysiological parallelism is neither necessary nor sufficient as a linking proposition. A major goal of their analysis is to evaluate the notion of a `bridge locus', which is essentially the `stage' in the conception of the `Cartesian theatre' in which percepts are brought to consciousness. They focus on the rich feedback interconnectivity throughout the cortex as a reason to doubt the existence of such a locus. But since the role of such feedback may well be limited to gain control (or `auto-ranging') functions, it may not interfere with an essentially feedforward notion of the information processing. If the feedback is there merely to keep the processed signals within the operating range at each successive stage, it should be regarded as part of the housekeeping functions rather than entering into the information-processing sequence.
What is more damning to the psychophysiological
parallelism, it seems, is the stunning multiplicity of such
representations (over thirty visual representation areas alone), in contrast
to the relatively unitary nature of perceptual experience. The problem
of identifying the homunculus in the audience of the Cartesian theatre
is that there are so many candidates. In one sense, each visual representation area is the audience for the Cartesian theatre of the previous one.
The brain is a veritable Broadway of competing shows, with each audience
performing on the stage of the next theatre. Is there a final arbiter
of the heterarchy of interacting critics?
Local specificity of neural representation
To propose, however, that this level of complexity precludes any
local specificity of perceptual processing seems
premature. Critics of fMRI studies often cite the complexity of the neural processing as an a priori reason that fMRI
data will not reveal anything useful about brain processing. The data
often support such criticisms, since they are typically obtained from
observers performing some discrimination task on complex stimuli, for which a substantial number of discrete brain areas are shown
to be activated.
But suppose you ask observers simply to look at well-defined stimuli and maintain a stable perceptual state? fMRI studies of responses to such stimuli, when compared with those for control stimuli differing on only one attribute such as coherent versus incoherent motion, typically show activation of only a single cortical representation area. Where did the whole brain interaction and multiple representation areas go in such cases? Critics argue that they are still present, but too weak to be seen at the available signal-to-noise ratio. In the limit it is impossible to argue against such an interpretation, but to a first approximation (say, at a 10 : 1 signal-to-noise ratio), such whole brain interactions are insignificant in relation to the primary signal.
What does it mean for the Cartesian theatre that a controlled stimulus difference may activate just one local brain area---a different one for each stimulus property? It suggests that the information for that stimulus property is stored locally, clearly eliminating the notion of a distributed representation. The mapping studies of large numbers of retinotopic visual maps already put a profound crimp in the distributed representation. Maps are themselves an explicit isomorphism. Although many authors agree with Dennett that isomorphism is not a necessary property of perceptual encoding, they need to deal with the fact that it is a physiological reality at many levels of analysis: anatomical, neurophysiological, neuropathological, and `fMRIical'.
Isomorphism may not be necessary, but it seems
to be useful judging by the number of brain maps that have been established. The brain has evolved to have multiple isomorphisms in many
sensory and motor systems. Their main utility is probably in minimizing
the length of wiring while maximizing connectivity. Just as it is usually
a lot easier to show someone how to construct a bookcase than to explain
it in verbal instructions, the intermediate step of an isomorphic mapping
may provide a basis for hypothetical manipulations that are not readily
accessible to a symbolic representation. The utility of such mapping seems
to have escaped critics of the Shepherdian concept of mental rotation.
This suggests the need for a new logical category: not necessary but efficient!
The fallacy of the personal
Finally, a commentary on the personal level of analysis. This
viewpoint was originally proposed as a way of identifying the agent responsible for actions in the world. As an analysis of perception, however, it
strikes me as incoherent. Typically, its proponents do not articulate
the process of perception when viewed from the perspective of the organism
in the environment. They simply assert that perception is
the organism interacting with the environment. From a solipsistic perspective, this viewpoint does not encompass the perception of which
I am predominantly aware in everyday life, and becomes, moreover, absurd
when considered in its logical extreme.
If perception is characterized as the interaction of the organism with the environment, then perception would have to be attributed to the following:
Amoebae and other single-celled organisms
Automated rapid-transit systems
Swimming-pool cleaners
Thermostats
Insects
Mars rovers
etc.
The only thing that in principle distinguishes the human interaction with the environment from these examples is that the human is a sentient, or conscious, person and we regard these others as not conscious (with perhaps a question mark over insects). Thus, the concept of the personal is simply an elaboration of the core element of the static perceiver; the complex person/environment system does not explain the perception of the static observer but, on the contrary, needs the element of static perception before it becomes a `person'.
Pessoa, Thompson, and Noe hold Marr guilty of the fallacy of the homunculus because he does not specify to `whom' the information about the primal sketch is provided. Presumably he has avoided the sin of isomorphism because the primal sketch is a symbolic description, rather than a remapping. But the fact that the information is provided to the next stage of neural processing does not imply the existence of a homunculus. I could, for example, propose a maximum spatiotemporal gradient detector operating on the 2 1/2-D sketch that would automatically drive eye fixation to the point in the field with the highest spatiotemporal contrast. Simple perceptual processing would have led to action in the environment without the intervention of a homunculus.
To extend the analogy, competing needs in the hypothalamus could automatically reweight the importance of objects processed at the point of regard, varying the fixation duration according to the significance of the information relative to the prevailing biochemical state of the blood. The actions are now modified according to a complex set of prior behaviors. Have I introduced a homunculus? No. I have merely begun to dissect the interacting components of neural processing of a highly complex network of brain nuclei. In doing so, I may have reduced the compelling mystery of human perception to a detailed biological enterprise. But that enterprise certainly begins with a set of isomorphic maps of the world in the brain, and probably even dynamic 3-D isomorphisms. Unless they are tracked step by step, philosophical arguments as to the necessity of their existence will remain a war of words.
Christopher W Tyler
Smith-Kettlewell Eye Research Institute, 2232 Webster St, San Francisco, CA 94115, USA.
Comments welcomed to: cwt@skivs.ski.org
Dennett D, 1991 Consciousness Explained (Boston, MA: Little, Brown)
Grossberg S, 1983 ``The quantized geometry of visual space: the coherent computation of depth, form, and lightness'' Behavioral and Brain Sciences 6 625--692
Marr D, 1982 Vision (New York: W H Freeman)
Pessoa L, Thompson E, Noe A, 1998 ``Finding
out about filling in: A guide to perceptual completion for visual
science and the philosophy of perception'' Behavioral and
Brain Sciences (in press)
© 1998 Pion Ltd