cutaneous color sensitivity: explanation and demonstration - American

Psychological Review

1966, Vol. 73, No. 4, 280-294

CUTANEOUS COLOR SENSITIVITY: EXPLANATION AND DEMONSTRATION W. L. MAKOUS 1 IBM Research Center, Yorktown Heights, New York Past reports that humans can, in complete darkness, sense with their fingers the colors objects would have if illuminated, have understandably been received with skepticism. A previously proposed hypothesis, based on differential rate of absorption of infrared radiation by different layers of the skin, is inconsistent with the thermodynamics of the situation. Quantitative analysis of the system consisting of a roomtemperature surface juxtaposed to the higher temperature skin, however, leads to the conclusion that large differences in emissivity of different room-temperature surfaces almost certainly could be detected by the associated effects on skin temperature. Easily repeatable demonstrations show that this is true, and a few simple temperature measurements confirm the theoretical identification of the mechanism.

Occasionally over the years popular accounts and clinical reports have appeared that describe a rather incredible capacity of some individuals to sense with their fingers the properties of objects that normally only the eyes can detect. Recent investigations, beginning with those of Goldberg (1963), Youtz (1963, 1964), and Nyuberg (1964), have increased the creditability of such reports to the point where they must be investigated to explain their apparent incompatibility with the present concepts of sensory function. Some of these reports are probably explained by the inadequacy of blindfolds for preventing use of the eyes (see Gardner, 1966). Among these is Nyuberg's (1964) report on an individual who identified printed characters and demonstrated trichromatic color sensitivity while blindfolded. In all respects, the subject's performance was remarkably like what one

would expect if it were mediated by images reaching the retina through the pinhole apertures that typically exist at the edges of blindfolds (Gardner, 1966). Goldberg (1963) reported earlier tests on the same individual under conditions that would seem to make peeking more difficult, but his report generally lacks signs of the rigor required in an investigation of unlikely phenomena. In such cases, where the results seem very unlikely, and especially in cases where a small number of individuals claim an extraordinary ability, rather extensive precautions against trickery may be required as is advised by Gardner (1966). In all cases in which light is intended as a cutaneous stimulus (e.g., in Leont'ev's work described by Pick, 1964, p. 22), interpretation of the results must take into account the thermal effects of light on skin; "heat filters" do not eliminate these effects, for no filter 1 The many helpful comments of colleagues can pass light and not pass energy and the technical assistance of Jean Kirby which can be converted to heat upon are gratefully acknowledged. The author's absorption. present addresses is Department of PsyYoutz's (1963, 1964, 1965) results chology, University of Washington, Seattle, are not so easily explained as those Washington. 280

CUTANEOUS COLOR SENSITIVITY mentioned above. He tested an exceptional individual who was able to identify the color of paper, plastics, and cloth, which were in the complete darkness of the interior of a light-tight box. (Color is used here in the technical sense, encompassing brightness as well as hue.) The subject was required to wear a blindfold and to rest her forehead against the box more than a foot above two holes in the side of the box, to which were attached thick, black sleeves with elastic wrist bands, through which the subject was required to thrust her hands in order to make the identifications. Four different series of experiments were conducted in this manner. The subject was correct 85-95% of the time in the first series, much less often but still significantly above chance in the second series, and at chance level in the third and fourth series. Youtz (1964, 1965) also tested 133 Barnard College freshmen under the following conditions: To each he attached a blindfold and also a "bib" stretching from around the subject's neck to the top of the frame holding the test objects, thus completely hiding the test objects from the subject. He reports that 10% of the subjects were able to perform under these conditions consistently (not Type I errors) above chance. To explain the performance of these subjects, Youtz has advanced what he calls the "wavelength-temperature discrimination hypothesis" (1964, 1965). Unfortunately, Youtz has not yet published a full account of this work so that his procedure and analysis can be carefully studied. Thus one could attribute the gradual loss of ability of his exceptional subject to learning on the part of the experimenter, and the probability statistics in his study of Barnard students are not reported in enough detail to make the data so

281

far reported thoroughly convincing. Yet, although the wavelength-temperature discrimination hypothesis is almost certainly incorrect, as shown below, the conclusion that humans can perform discriminations of the type described by Youtz almost certainly is correct. The purposes of this paper, then, are: 1. To point out the flaws in the wavelength-temperature discrimination hypothesis. 2. To show that we should not be surprised if people can discriminate objects in the dark, which appear to differ only in color when illuminated; however, certain thermodynamic considerations and established sensory capacities would lead us to be surprised if people could not make such discriminations. 3. To describe a simple demonstration whereby the skeptical reader can prove to himself his ability to make such discriminations. WAVELENGTH-TEMPERATURE DISCRIMINATION HYPOTHESIS The inevitable consequence of absorption of radiant energy, whether in the visible spectrum or not, is an increase in temperature of the absorbing entity. We know that the human skin is exquisitely sensitive to temperature changes: Under some conditions a human subject can detect a cutaneous temperature change of .001 °C sec-1 (Hardy & Oppel, 1937; Hendler & Hardy, 1960; Hendler, Hardy, & Murgatroyd, 1961; Hensel, 1950). Thus it is of no special interest that a subject can detect the differences caused by illumination of the skin by different light colors unless the difference in energy absorbed is smaller than is likely to be detected

282

W. L. MAKOUS

by the resulting effects on skin temperature, a capacity that has not yet been convincingly demonstrated. Discrimination of objects that are neither illuminated nor self-luminous poses a different problem, however, for the way in which the object serves as an energy source may not be obvious. To account for discrimination of such objects, Youtz (1964, 1965) has advanced the following hypothesis. The human skin, a nearly ideal black-body radiator under ordinary conditions (Barnes, 1963; Hardy, 1934b; Hardy & Muschenheim, 1934), radiates energy within the dark, light-tight box containing the test objects. The energy radiated by the skin is absorbed by the test objects and reemitted at wavelengths determined by their respective compositions, particularly by any dyes or pigments present. The rays reemitted from the test objects penetrate the skin to varying depths before being absorbed, depending upon their wavelengths (Hardy & Muschenheim, 1934, 1936; Oppel & Hardy, 1937a, 1937b). Localized heating then occurs at the depths at which the rays are absorbed, and the subject hypothetically discriminates the relative depths at which the heating occurs. Necessary for this hypothesis, then, are: (a) detectable radiation by the test objects in wavelengths which penetrate skin to different depths, and (6) the human ability to discriminate slight differences in temperature, or differences in temperature change, at different depths within the skin. This hypothetical explanation is plausible enough. The lack of independent evidence that humans can make the necessary b discrimination is only a minor drawback. It is shown below, however, that the hy-

pothesis fails to satisfy condition a; for, under the conditions of Youtz's experiments, practically all of the energy is radiated by the test objects in wavelengths that are absorbed too close to the skin surface to provide depth cues. Thermodynamic Considerations The foregoing conclusion follows from certain thermodynamic considerations that hold for all but very special conditions such as are artificially arranged within the laboratory or are inferred to exist in exotic places like the interiors of stars. Since the hypothesis concerns only radiant heat exchanges, discussion of the effects of conduction and convection will be deferred until later. The principal point here is that a substance of a given temperature, such as the subject's skin or one of the test objects, cannot radiate more energy at any wavelength than does an ideal radiator at the same (or higher) temperature, unless it receives energy from a higher energy source. If the substance in the lower energy state (the test object) receives energy from the substance in the higher energy state (the skin) only through radiation, to which the present case is limited, then the energy absorbed cannot be reemitted (fluoresced) at a shorter wavelength than the wavelength at which it was originally absorbed (Stokes1 law). This means that so long as the test objects are cooler than the skin, they always return less energy to the skin in wavelengths shorter than any particular wavelength than the skin emits in those wavelengths; this, then, places an upper limit on the amount by which test objects can differ from one another in their radiant exchanges with the skin. These limits will now be taken up quantitatively.

CUTANEOUS COLOR SENSITIVITY The proportion of incident radiation absorbed rather than reflected by an opaque object at a given wavelength is expressed by its spectral emissivity, ex. A perfect reflector reflects all incident radiation and therefore absorbs none (ex = 0): An ideal blackbody radiator absorbs all incident radiation (ex6 = 1.0), but it emits energy according to Planck's radiation formula: s.\-f>

H\. =

where H\ = spectral radiant flux density in w cm^1 mpr1, X = wavelength in /it, T = temperature in K°, d = 37.403 w cm-1 niAr1, c* = 14387.9 M° K. The energy radiated by the skin at a given wavelength, H\t, equals the product of its spectral emissivity, ex,, and the energy radiated at that wavelength and temperature by a black body, #x6:

rr -

•"X,

rr

*Xr"Xi, —

ex/iX~6 _

Then the maximum amount of energy of wavelength X that could be returned to the skin by a juxtaposed test object, if it were a perfect reflector, is the total amount emitted by the skin, H\,. (To simplify geometric considerations, the distance of the test object from the skin is assumed to be small relative to its length and width.) And the least amount of energy of wavelength X that could be returned to the skin by a juxtaposed test object, if it were an ideal radiator, is H\b, of which would be absorbed by the skin. The net energy loss of the skin over the interval, X = a to X = z, is the difference between the energy radiated and the energy absorbed over that interval. The least that loss can be,

283

when the skin is juxtaposed to a perfect reflector, is zero; and the greatest it can be, when juxtaposed to an ideal radiator, is:

Then the interval, AUmax, is the maximum range over which the test objects can vary in their effects on the skin. From this equation it is clear that this range depends upon the spectral emissivity of the skin, ex,; upon the temperature of the skin, T,; upon the temperature of the test objects, Tt,; and upon the spectral band considered, a < X < z. Computations Radiation in different wavelengths can be classified according to whether skin is relatively transparent to it, relatively opaque to it, or of intermediate density with respect to it. These three categories partition the relevant part of the spectrum into three regions (Hardy, Hammel, & Murgatroyd, 1956; Hardy & Muschenheim, 1934,1936; Hendler & Hardy, 1960; Hendler et al., 1961; Oppel & Hardy, 1937a, 1937b): Wavelengths that are, respectively, shorter than .8 p, longer than 3 n, and between .8 n and 3 M- For expository purposes, and because Youtz has implicated these three regions in his hypothesis, each will be considered separately. It is clear at the outset that practically no energy is emitted by the skin in the wavelengths shorter than .8 /t, that is, a* 0,

for the eyes are so sensitive to these

284

W. L. MAKOUS

wavelengths that we should otherwise be able to see one another in the dark. To bias the result in favor of the wavelength-temperature discrimination hypo thesis ,^let «n. and T, assume their maximum respective values of 1.0 and 98.6°F and let, 0.

Then,

AH, = -1.13(10~22) w cm (The negative sign indicates that net energy flux is outward from the skin.) Since the energies of photons in these wavelengths exceed 2.4 X 10~19 j/ photon, the ratio, -1.13(10-") wcm- 8 2.4(10-19)j/photon = 4.7 (10~4) photon sec"1 cm~2 = 1.7 photon hr"1 cm~2,

be a room temperature of 72 °F. Then

+ 6

= -7.5(10- )+3.3(10-6) = — 4.2(10-") wcm- 2 = — 10~6 g-cal sec-1 cm"2. The threshold radiant flux density for large skin areas, -Hth«.h, depends upon a number of variables, but in this situation it lies between 1.5 X 10~4 and 7.0] X 10"> g-cal sec"1 cm"2 (Hardy & Oppel, 1937; Hendler & Hardy, 1960 ; Hendler et al., 1961 ; Lele, Weddell, & Williams, 1954). This is greater by two orders of magnitude than AHaun., the maximum possible energy exchange that could take place between the skin and a room-temperature object via wavelengths shorter than 3 jj. One might argue that less energy is required to reach threshold at these wavelengths than at the longer wavelengths used in arriving at the above value of Hthnah, but just the opposite appears to be the case (Oppel & Hardy, 1937a). If, nevertheless, there is some error in the conclusion that the signal afforded by 10~6 g-cal sec"1 cm-2 is too weak to be detected, then under the conditions here this signal would almost certainly be lost in the noise associated with the much larger energy exchanges occurring via the longer wavelengths (X > 3 ;u). These exchanges,

shows that a square centimeter of skin at body temperature loses, on the average, no more than two photons in these wavelengths every hour. Consider now radiation in the wavelengths between .8 p and 3 n, which penetrates to intermediate depths below the skin surface. The spectral emissivity of the human skin in these wavelengths is reported to be between .75 and .95 (Barnes, 1963; Hardy, 1934b; Hardy & Muschen= - / flx/*A + I heim, 1934; Oppel & Hardy, 1937a), Js Ja but again to bias the result in favor = 9.3 (10-8) wcm- 2 of the wavelength-temperature dis= 2.2 (10-3) g-cal sec-1 cm-2, crimination hypothesis, let «\, and T, assume their respective maximum are more than 2000 times those ocvalues of 1.0 and 98.6°F, and let Tb curring in the shorter wavelengths just

CUTANEOUS COLOR SENSITIVITY

285

discussed. Computations show that distant from receptors. Thus, the variations of ±.013°F in Tt, or Ts, necessary condition is not satisfied, or variations of ±.001 in emissivity and the wavelength-temperature diseither of the skin or of the test objects, crimination hypothesis cannot hold. would cause variations in the value of Mechanism Af/max for the wavelengths longer than Despite the flaws of the wave3 /i, as large as the entire value of length-temperature discrimination hyA.ffmax for the wavelengths shorter pothesis, it has led to an understandthan 3 n. Then no detectable radiant ex- ing of the situation it was intended to change occurs via wavelengths shorter explain. The total radiant heat than 3 #, and if depth of penetration exchange between the skin at body at provides the subjects with any in- temperature and a black body 2.2 X 10~3 g-cal formation, it must occur via wave- room1 temperature, 2 lengths longer than 3 ju. But all sec" cm~ , is anywhere from 3-15 wavelengths longer than 3 p are ab- times threshold, and thus it is cersorbed at approximately the same tainly large enough to be detected. depth (Hardy & Muschenheim, 1934). Objects having sufficiently different Furthermore, 90% of this radiation is spectral emissivities in the infrared, absorbed within the outermost 30 n when held close to the skin, should or so of skin (Hardy & Muschenheim, produce detectably different rates of 1936), which is keratinized tissue de- skin cooling, so long as a sufficient void of nerve endings and receptors temperature difference is maintained (Ranson & Clark, 1953). Little more between them and the skin. That than 1% of incident radiation in these subjects use this temperature cue as wavelengths penetrates even so far as the basis of their discriminations is the most superficial skin layers sus- supported by some of their subjective pected of harboring thermal receptors, reports (Youtz, 1964). One can, in fact, compute the more than 100 n below the surface (Hendler et al., 1961; Zotterman, differences in emissivity that should 1959). That radiation in these wave- be detectable on this basis with any lengths does not penetrate any signifi- given combination of temperatures. cant distance is supported by the Let real substances, having spectral finding (Hendler et al., 1961) that emissivity, ex,, replace the ideal subblackening the skin surface, which stances in the equation for A£fmi>x, and insures complete absorption of inci- let ex, = 1.0, a = 0, and z = oo, then dent radiation at the very surface of the net energy loss by the skin is the the skin, has no effect upon thresholds difference between the total energy emitted by the skin and the sum of to such radiation. To recapitulate, then, a necessary energy emitted and the energy recondition for the temperature-wave- flected by the substance, all of which length discrimination hypothesis is is absorbed by the skin (because that the test objects emit detectable ex. = 1.0). Thus, radiation at wavelengths that penetrate to different depths within the X Jo ' Jo skin. But, as just demonstrated, practically all of the energy is exchanged via wavelengths that are absorbed at a uniform depth, and that depth is very close to the surface but Jo

r

r

-r

286

W. L. MAKOUS

The solid lines in Figure 1 show the effects of skin temperature, T,, on the likely upper and lower limits of Ac