Recent work in philosophy of psychology has been suspicious of the mental in general and of phenomenology in particular. It is widely assumed (1) that phenomenology does not and should not play a substantial role in a scientific psychology, and (2) that mentalistic notions appear in psychology only as theoretical posits. This paper examines work in the psychophysics of vision (the Weber-Fechner laws, the Craik-O'Brien-Cornsweet effect and subjective contour figures) and argues against these two claims on the grounds that psychophysics, commonly recognized as among the most "scientifically mature" areas of psychology, has a strong methodological and ontological commitment to phenomenology, and that the phenomenology of percepts comprises data for theoretical psychology rather than theoretical constructs.
The history of philosophy of mind in the twentieth century has been in no small measure a story of suspicion towards mentalistic categories in general and to the first-person, experiential, phenomenological character of the mental in particular. It has been argued variously that mental states do not exist at all, that they are methodologically unacceptable for a scientific psychology, that they are identical with brain states or behaviorial dispositions and that they are causally inert epiphenomena. And such claims have been advanced on grounds of methodology, of metaphysics and of an analysis of the history of science. It is widely believed on the current scene that mental states need to be "naturalized" if they are to appear in a scientific psychology or a serious metaphysics. Thus recent disputes between reductionists and eliminativists are carried on within the shared view that the mental needs to be grounded in something other than its phenomenology if we are to have it at all.
I believe that this conversation is built upon several very bad assumptions. The first assumption is that the areas of psychology that are generally deemed to be most scientifically respectable (notably, psychophysics) are not tied to phenomenological features of the mental. The second is that mentalistic notions appear in psychology only as theoretical posits. I believe that both of these assumptions are wrong. In point of fact, a significant portion of psychophysics is very much in the business of describing relations between phenomenological properties (percepts) and non-phenomenological properties. And since psychophysics supplies much of the data that theoretical psychology attempts to explain, phenomenologically-described mental states make up much of the data of psychology, and not merely its theoretical posits. And hence the evidential status of these mental states is independent of the status of any truly theoretical mental states (e.g., infra-scious beliefs and desires) posited as part of a retroductive explanation.
Psychophysics and Scientific Psychology
While there are many areas of psychology whose status as science are often called into doubt, the main exception to this suspicion is the kind of experimental psychophysics that was pioneered around the turn of the century by figures such as Fechner, Weber, Mach and Helmholtz. I shall discuss three examples of psychophysical data from the vision literature: the Weber-Fechner Law, the Craik-O'Brien-Cornsweet effect, and the Kanizsa square. These three examples will illustrate the points, respectively, (1) that psychophysics deals with relationships between stimuli and "subjective" phenomenological properties, (2) that in some cases it is very much the qualitative properties of mental states that are the subject matter of psychophysics, while (3) in others, intentional properties also seem to play a major role.
The Weber-Fechner Law and Phenomenology
The Weber-Fechner law is perhaps the best known result from nineteenth-century psychophysics. Its general claim is that for the various perceptual modalities, the intensity of the percept is a logarithmic function (alternatively, a power function) of the intensity of the stimulus. In the case of vision, for example, this law relates differences in the apparent brightness of a figure&emdash;how bright it seems to an observer&emdash;to differences in the absolute luminance of the stimulus (how much light is really reflected from it). I shall follow the practice of refering to the experiential property of the percept as brightness and the objective property of the stimulus as luminance. Counter-intuitively, the relationship of brightness to luminance is not linear but logarithmic.
Now what is the Weber-Fechner law about? It is a description of a function from stimulus intensity (luminance) to percept intensity (brightness). Brightness, however, is a phenomenological property And, more generally, to call a thing a percept is to describe it in phenomenological terms. But if the Weber-Fechner law is a paradigm example of scientific psychophysics, and if its subject matter involves a phenomenological property, then scientific psychophysics includes phenomenological properties in its domain of discourse.
The Craik-O'Brien-Cornsweet Effect
The same point can be made by consideration of the description of visual effects, such as the Craik-O'Brien-Cornsweet effect (COCE). The COCE involves two adjacent figures that are identical in luminance profile (i.e., in distributions of absolute measurements of reflected light) but differ in brightness (i.e., in the subjective perception of lightness and darkness). There are several ways of inducing this effect. The classic way is for the figures to be of a constant level of luminance except for the region very close to their border, with a slight increase or cusp on one side of the border and a slight decrease on the other:
The resulting percept is one of two figures of different brightness, each of which appears to be of constant brightness internally. The percept is, indeed, much the same as what would be produced by setting two figures of different luminance levels side by side so that the luminance profile is step-shaped:
In layman's terms, the two regions are identical in terms of the objective property of luminance profile, but one looks darker than the other. The difference in brightness between rectangles depends upon the difference in luminance at the borders.
Effects such as COCE present problems which it is the business of theoretical work in vision to solve. The problem, in this case, is a mismatch between the stimulus and the percept: local differences in brightness in the percept do not correspond to differences in luminance in the stimulus. Thus this kind of effect provides a kind of black box description of a function from a stimulus (in terms of a pattern of luminance that stimulates the retina) to a percept (in terms of an image that has contrasts in perceived brightness). Any viable model of the human visual system should be constrained by such descriptions, in the sense that their output should correspond to the percept when their input corresponds to the stimulus.
Again, the datum presented by this effect and to be explained by a theory of vision is a relation between phenomenological properties (how things look) and physical properties (how the patches reflect light). The reason it counts as a psychological effect is because the curve describing the brightness profile of the percept does not match the curve describing the luminance profile of the stimulus. But for our philosophical purposes what is important is that one of the curves reports purely phenomenological features. First, there is just no way around the fact that what is reported in this effect is that one patch looks brighter than the other, even though there is no difference in luminance. And it is hard to see how "looking brighter" can be anything other than a comparison in terms of phenomenological properties. (It cannot, for example, be paraphrased in terms of detecting differences in luminance, because in this case those differences do not exist.) There is no other way to get at brightness as a datum other than by examining your own percepts or accepting other people's reports of their percepts. And indeed, most effects of this kind are sufficiently reliable across subjects that reports of perceptual effects can be accepted on the basis of a very small test group. (In fact, one generally feels confident generalizing from one's own case.)
Second, it is the properties of the percept that supply the success conditions for an adequate theoretical model of the effect: a good model is one whose output matches the percept and has a plausible neural interpretation. But note that the effect enters the literature as a datum without such a theoretical explanation, and is not imperiled as a datum in the absence of theoretical explanation or neural correlation. Indeed, it is the data that constrain the model and the localization, and not vice-versa. And hence you simply cannot banish the qualitative aspect of such effects from your description of the psychophysical data: eliminate the qualitative phenomenological property of percept brightness and you have not sanitized the portion of psychophysics concerned with brightness, but eliminated it entirely. No phenomenology, no psychophysics.
Indeed, discussions of theoretical work in perception sometimes turn precisely upon the question of whether a given model explains the percept. For example, some researchers have suggested that the COCE is explained by the fact that luminance profiles of steps and cusps have similar abstract properties. Todorovic (1987) criticizes the theories cited on the grounds that they do not account for the appearances:
The problem is that there is a mismatch between the shape of the brightness profile of the percept and its presumed neural counterpart ....The luminance cusp distribution ...gives rise to a percept that has the shape of a step ....However, the presumed physiological foundation of the percept, according to the preceding analysis, has a quite different profile, one that is more similar to the cusp-shaped profile of the underlying luminance distribution. (page 547, emphasis added)
The issue here is quite clear: it is not enough for a theory of vision to accommodate the neural data. It must accommodate the phenomenological data&emdash;the "appearance of the stimulus"&emdash;as well. Thus theorists such as Todorovic clearly regard the phenomenology of vision as setting important constraints on what can count as a successful visual theory. It appears that researchers in cognitive neuroscience place more weight upon phenomenology than do many contemporary philosophers.
The Kanisza Square and Intentionality
The COCE illustrates the fact that qualitative phenomenological properties such as intensity of qualia are often essential to psychophysical data. There are also psychophysical data in which at least simple intentional properties seem to be essential. A visual effect that illustrates the importance of a minimal form of intentionality is the Kanisza square (Kanisza, 1979), depicted below:
In viewing this figure, normal subjects report seeing a square that is slightly brighter than the background. The subject thus "perceives" boundaries corresponding to the sides of a square&emdash;boundaries that are not "really there" in the sense that there is no discontinuity in luminance in the portions of the stimulus where boundaries are perceived. Normal perceivers also perceive the interior of the square as slightly brighter than the background, although in fact there is no difference in luminance between interior and background regions. Here the visual system is somehow "filling in" boundaries that are not there to be seen and producing an interpretation of the brightness of the interior region of the figure it supposes to be there. In lay terms, we "see a figure that isn't there" and see it as being "brighter than it should be."
Here, again, there is a well-defined difference between the phenomenology of the percept and the gross physical properties of the stimulus. The percept has a figure, borders and brightness contrast that are nowhere to be found in the stimulus. The constraint such an effect places upon theoretical work in vision is, again, that one's model of the visual system ought to reproduce the psychophysical phenomena observed in human subjects. A model whose output represents the interior of the "square" and the background as of the same brightness, or which does not represent boundaries along the "sides" of the "square", or which does not pick out a square at all, is not an adequate explanation of the psychophysical data, because the output of the model does not correspond to the percept.
This effect, like the COCE, involves qualitative phenomenological properties such as the brightness profile of the percept. However, this example also involves something not found in the previous examples: the perception of a figure as such. In the previous effects, we had discrete regions that could be isolated objectively both spatially and in terms of luminance profile. In this example, however, the subjects "sees" a square some of whose boundaries are not marked by any objective properties. Here we have a Gestalt phenomenon in which one "constitutes" a region as a figure of a given kind. The subject "sees" this region "as a square", and indeed as a square that is brighter than its background. This kind of Gestalt phenomenon is a very simple case of intentionality. It involves seeing a region as a figure of a given kind, and seeing-as is intentional in nature. Moreover, it also bears that feature of intentionality emphasized by Brentano (1874) and Chisholm (1957): namely, the fact that there is an "intentional object" (the percept of a square) to which nothing objective need correspond. (And indeed in this case there is no square that corresponds to the percept.)
Now this kind of Gestalt phenomenon is every bit as interesting a psychological datum as are the purely qualitative properties that appeared in Weber-Fechner and the COCE. And there is indeed some reason to think that any theory of the qualitative effects cannot be done independently of this kind of simple figure-constitution. There is evidence, for example, that the visual system is relatively insensitive to gradients of luminance within the boundaries of a figure, and that it "fills in" the interior of a figure. This would indicate that constitution of figures is not simply a later stage of cognition that takes a pre-given qualitative input, but rather that there is significant interaction between the factors that produce the perception of boundaries and those that produce qualitative features such as brightness. There are likewise questions about whether the perception of boundaries in such cases is completely independent of the perception of figures (arguably a primitive intentional state), and hence about whether one can fully isolate processes of "edge detection" from intentionality.
The moral, again, is that phenomenological properties figure significantly in our psychophysics, and our psychophysics is what provides the data for (and hence the constraints upon) our theoretical psychology of perception. In this case, it is not only qualitative phenomenological properties, but intentional properties. (There is a "what-it's-like" to seeing something as a square, and it is different from the "what-it's like" of seeing something as a triangle or simply having sensations.) You cannot throw out the phenomenology and keep the data, because the data relate phenomenological properties to physical properties.
These three examples are designed to give the reader some sense of the kinds of data collected in psychophysics of vision. They are representative of much research in experimental psychology, which involves the collection of "effects" that would need to be explained by a theory of perception. Indeed, it would seem that phenomenology plays at least four distinct roles in psychophysics.
(1) The subject matter of psychophysical phenomena involves phenomenologically-described mental states. More precisely, psychophysical data like those described above treat the visual system as a function from objectively-described stimuli to phenomenologically-described percepts.
(2) First-person phenomenological description is vital to the description of psychophysical data. If what we are after is, say, an explanation of how things look (say, the fact that the Kanisza figure looks like a square that is brighter than its background), it is hard to see how to describe what we want to explain in non-phenomenological terms. While it is indeed desirable to seek a neural correlate to the percept, it is the phenomenologically-described percept that provides the constraints necessary for judging whether a given neural phenomenon has the right properties to serve as such a correlate.
(3) The reliance upon phenomenological data does not result in any perilous unreliability or problems of confirmation. Indeed, most psychophysical data of this sort are remarkably stable across human perceivers, and experimentalists in vision tend to regard their own percepts as having significant intersubjective validity.
(4) The phenomenology of the percept is in fact central to the methodology of researchers in psychophysics. The best evidence for this claim I have encountered is anecdotal. When researchers in psychophysics of perception present papers at their professional meetings, I am told, a great deal of care is lavished upon producing the best possible visuals&emdash;i.e., visuals that allow the audience to experience the effect for themselves. Indeed, I am told that audiences tend to be impatient with data plots and care principally about their ability to "see" the effect. The primary validation of the effect comes through the researcher's own experience of the percept.
Phenomenology, Intentionality and Psychology's Data
The foregoing discussion of psychophysics of vision does much to belie the current wisdom about the role played by subjective mental states in psychology. First, consider claims that a scientific psychology should not be committed to mental states&emdash;or at least to mental states characterized in a way that is dependent upon their phenomenology. Psychophysics is widely regarded as the portion of psychology that really has become scientific, and it depends very heavily upon phenomenology. On the one hand, its domain includes phenomenologically-described mental states (percepts). On the other hand, its methodology requires subjective access to the first-person, experiential, phenomenological character of these percepts. And without such a phenomenoogically-based psychophysics there can be no theoretical psychology of perception, because there will be no data for it to explain! Moreover, psychophysics is in no way apologetic about its references to percepts. There is no suggestion that they are in need of "vindication" by way of unification of psychology with a larger naturalistic body of science, or even with neuroscience. Psychophysics treats percepts (or better, relations between objective properties of stimuli and properties of percepts) as its domain without apology. And it is the psychophysics that holds theoretical psychology&emdash;including neuroscience&emdash;to the test, and not the other way around. A neuroscientific theory that fails to duplicate the psychophysics of human perception is an inadequate theory of perception. Given a mismatch between psychological theory and psychophysical data, it is theory that is regarded as suspect.
This may not be enough to vindicate the mental or its phenomenological aspects to a diehard anti-mentalist. But it does seem to present a set of options starker than those generally proposed. We can, on the one hand, embrace psychophysics, and with it the phenomenology of perception. Or we can reject psychophysics along with all the rest of the mental. The result of the latter course, however, is not a naturalistic psychology, but no psychology at all. There can be no theoretical psychology of phenomena like vision&emdash;be it intentional, computational, connectionist or neuroscientific&emdash;because there is no longer a domain of data for such theories to explain. There can, of course, be experimental data about things like firing potentials and receptive fields and anatomical data about things like projections of fields of cells, but this does not add up to a psychology, because the psychologically relevant functional units in the nervous system can only be inferred from (indeed can only be constituted in terms of) the tasks they perform, and in order to have a demarcation of those tasks we must rely on data from psychophysics. Far from psychology being "displaced by a mature neuroscience", there would be no neuroscience beyond the level of physiology. It is one thing to say that psychological data present puzzles that we may not be able to find solutions to, or solutions of a particular kind. It is quite another to say that the data do not exist.
Second, these examples belie the claim that the only role played by mental states is as "theoretical posits" of psychology. This issue has become unduly confused due to the fact that much talk about mental states as "theoretical entities" trades upon an equivocation. "Seeming brighter" may be "theory-laden" in the bland sense that accrues to all terms if one accepts the network theory of meaning, but it is surely not "theoretical" or "retroductive" in the strong sense that neutrinos are "theoretical" or that Pluto was "theoretical" before telescopic confirmation of its existence. While some "mental states" may be truly theoretical, qualia and perceptual Gestalten are not. They are the data of theoretical psychology, not its explanatory posits.
What I have attempted to argue in this paper is that even a brief examination of psychophysics will reveal as erroneous two key assumptions of much of the contemporary debate about the nature of the mental and the shape of a scientific psychology. First, the best-established part of scientific psychology is essentially committed to phenomenological properties of mental states, both as its domain and as a necessary part of its methodology. Second, the mental states that appear in psychophysics under phenomenological descriptions appear not as the posits, but as the data for theoretical psychology, and thus are not so readily subject to elimination as a consequence of theory change. Indeed, psychophysics provides both the problems that theoretical psychology needs to solve and the constraints within which one can offer a realistic theory of human cognition.
Arend, L.E. (1973) Spatial differential and integral operations in human vision: Implications of stabilized retinal image fading. Psychological Review, 80, 347-395.
Békésy, G. von (1972) Compensation method to measure the contrast produced by contours. Journal of the Optical Society of America, 62, 1247-1251.
Benary, W. (1924) Beobachtungen zu einem Experiment über Helligkeitskontrast. Psychologische Forschung, 5, 131-142. [Translated as "The influence of form on brightness contrast,"" In W.D. Ellis (ed.) A source book of Gestalt psychology. London: Routledge & Kegan Paul, 1938.]
Brentano, F. (1874) Psychologie vom empirischen Standpunkt. Duncker & Humbolt.
Campbell, F.W., Howell, E.R., & Johnstone, J.R. (1978) A comparison of threshold and suprathreshold appearance of gratings with components in the low and high spatial frequency range. Journal of Physiology, 284, 193-201.
Chisholm, R. (1957) Intentional Inexistence. In Perceiving: a Philosophical Study, by Roderick Chisholm. Ithaca: Cornell University Press.
Cohen, M.A. & Grossberg, S. (1984) Neural dynamics of brightness perception: Features, boundaries, diffusion, and resonance. Perception & Psychophysics, 36, 428-456.
Cornsweet, T.N. (1970) Visual Perception. New York: Academic Press.
Craik, K.J.W. (1940) Visual adaptation. Unpublished doctoral thesis, Cambridge University.
Davidson, M. & Whiteside, J.A. (1971) Human brightness perception near sharp contours. Journal of the Optical Society of America, 61, 530-536.
Dooley, R.P. & Greenfield, M.I. (1977) Measurements of edge-induced visual contrast and a spatial-frequency interaction of the Cornsweet illusion. Journal of the Optical Socieity of America, 67, 761-765.
Ehrenstein, W. (1974) Über Abwandlungen der L. Hermannschen Helligkeitserscheinung. Zeitschrift für Psychologie, 150, 83-91.
Fechner, G.T. (1877) In Sachen der Psychophysik. Leipzig: Breitkopf & Härtel.
Fechner, G.T. (1882) Revision der Hauptpuncte der Psychophysik. Leipzig: Breitkopf & Hartel.
Gerrits, H.J.M. & Vendrik, A.J.H. (1970) Simultaneous contrast, filling-in process, and information processing in man's visual system. Experimental Brain Research, 11, 411-430.
Gilchrist, A. (1977) Perceived lightness depends on perceived spatial arrangement. science, 195, 185-187.
Grossberg, S. (1983) The quantized geometry of visual space: The coherent computation of depth, form, and lightness. Behavioral & Brain Sciences, 6, 625-692.
Grossberg, S. (1987a) Cortical dynamics of three-dimensional form, color, and brightness perception, I: Monocular theory. Perception & Psychophysics, 41, 87-116.
Grossberg (1987b) Cortical dynamics of three-dimensional form, color, and brightness perception, II: Binocular theory. Perception & Psychophysics, 41, 117-158.
Grossberg, S. & Mingolla, E. (1985a) Neural dynamics of perceptual grouping: Textures, boundaries, and emergent segmentations. Perception & Psychophysics, 38, 141-171.
Grossberg, S. & Mingolla, E. (1985b) Neural dynamics of form perception: Boundary completion, illusory figures and neon color spreading. Psychological Review, 92, 173-211.
Grossberg, S. & Mingolla, E. (1987) Neural dynamics of surface perception: Boundary webs, illuminants, and shape-from-shading. Computer Vision, Graphics & Image Processing, 37, 116-165.
Growney, R.L. & Neri, D.F. (1986) The appearance of the Cornsweet illusion: Measures of perceived contrast and evenness of brightness. Perception & Psychophysics, 39, 81-86.
Grüsser, O-J. (1993) The discovery of the psychophysical power law by Tobias Mayer in 1754 and the pscychophysical hyperbolic law by Ewald Hering in 1874. Behavioral and Brain Sciences Vol 16, No. 1: 142-144.
Gundlach, H. (1993) Psychophysics, its history and ontology. Behavioral and Brain Sciences Vol 16, No. 1: 144-145.
Hamada, J. (1982) The contour-enhancement effects produced by darkening effects. In H.-G. Geissler & P. Petzold (eds.), Psychophysical judgment and the process of perception. Amsterdam: North-Holland.
Hamada, J. (1984) A multi-stage model for border contrast. Biological Cybernetics, 51, 65-70.
Hamada, J. (1985) Asymmetric lightness cancellation in Craik-O'Brien patterns of negative and positive contrast. Biological Cybernetics, 52, 117-122.
Heggelund, P. & Kreklink, S. (1976) Edge dependent lightness distributions at different adaptation levels. Vision Research, 16, 493-496.
Heinemann, E.G. (1972) Simultaneous brightness induction. In D. Jameson & L.M. Hurvich (eds.) Handbook of sensory physiology: Vol VII/4. Visual Psychophysics. Berlin: Springer-Verlag.
Helson, H. (1963) Studies of anomalous contrast and assimilation. Journal of the Optical Society of America, 53, 179-184.
Horst, S. (1995) Eliminativism and the Ambiguity of 'Belief'. Synthese 104, 123-145.
Isono, H. (1979) Measurement of edge-induced visual contrast (Note No. 233) NHK Laboratories.
Kanizsa, G. (1979) Organization in vision. New York: Praeger.
Katz, D. (1935) The world of colour. London: Kegan Paul, Trench, Trubner.
Koffka, K. (1935) Principles of Gestalt Psycohlogy. New York: Harcourt, Brace.
Krauskopf, J. (1967) Heterochromatic stabilized images: A classroom demonstration. American Journal of Psychology, 80, 634-636.
Kries, J. von (1882) Über die Messung intensiver Grössen und über das sogenannte psychophysische Gesetz. Vierteljahrsschrift für wissenschaftliche Philosophice 6:257-94.
Mayer, T. (1754) Observationes astronomicae A. 1853 Gottingae habitae. Commentarii Societatis Regiae Scientiarium Gottingensis 3:444-51.
Mayer, T. (1755) Experimenta circa visus acienm. Commentarii Societatis Regiae Scientiarium Gottingensis 4:97-112.
McCourt, M.E. (1982) A spatial frequency dependent grating-induction effect. Vision Research, 22, 119-134.
O'Brien, V. (1958) Contour perception, illusion, and reality. Journal of the Optical Society of America, 48, 112-119.
Plateau, J.A.F. (1872) Sur la mesure des sensations physiques, et sur la loi qui lie l'intensité de ces sensations a l'intensité de la cause excitante. Bulletins de l'Académie Royale de Belgique (2nd series) 33:376-88.
Quinn, P.C. (1985) Suprathreshold contrast perception as a function of spatial frequency. Perception and Psychophysics, 38, 408-414.
Ratliff, F. (1965) Mach bands: Quantitative studies on neural networks in the retina. San Francisco: Holden-Day.
Ratliff, F. (1978) A discourse on edges. In J.C. Armington, J. Krauskopf & B.R. Wooten (eds.), Visual psychophysics and physiology, New York: Academic Press.
Ratliff, F. & Sirovich, L. (1978) Equivalence classes of visual stimula. Vision Research, 18, 845-851.
Spillman, L. & Levine, J. (1971) Contrast enhancement in a Hermann grid with variable figure-ground ratio. Experimental Brain Research, 13, 547-559.
Stevens, S.S. (1951) Handbook of experimental psychology. Wiley.
Stevens. S.S. (1975) Psychophysics: Introduction to its perceptual, neural and social prospects, ed. G. Stevens. Wiley.
Todorovic, D. (1987) The Craik-O'Brien-Cornsweet effect: New varieties and their theoretical implications. Perception & Psychophysics, 42, 545-560.
White, M. (1979) A new effect of pattern on perceived lightness. Perception, 8, 413-416.