Psychophysical measurement of spectral sensitivity and color vision in red-light-reared tree shrews (Tupaia belangeri)

Vision Res Vol.31,No. 10,pp.174%1757, 1991 Printed inGreat Britain. All rights reserved

PSYCHOPHYSICAL MEASUREMENT OF SPECTRAL SENSITIVITY AND COLOR VISION IN RED-LIGHT-REARED TREE SHREWS (TUPAIA BELAiVGERI) HEYW~~D M. PETRY* and JOHNP. KELLY apartment

of Psychology, State University of New York at Stony Brook, Stony Brook, NY 11794, U.S.A.

(Received 3 October 1990; in revised form 4 February 1991)

Abstract-The role of the spectral lighting environment on the post-natal developmentof’ SpeCtrai sensitivity and color vision was studied in tree shrews (Tupaia belangeri) that were born and rearedto adulthood in cyclic red light. Normal tree shrews are dichromats, possessing short-wavelength-sensitive (SWS) and long-wavelength-sensitive (LWS) cone receptors and a small population of rods. Red-lightrearing (RLR) produced differential stimulation of the cone types by effectively eliminating photic stimulation of the SWS cones, without depriving the LWS cones. Spectral sensitivity and color vision were measured behaviorally for RLR shrews and normal shrews under different ambient light levels. Spectral sensitivity functions were deutan-like, exhibiting maxima at cu 450 and 550 nm and a minimum at 510 nm. No sibilant differences in spectral sensitivity were observed between RLR and control animals. Furthermore, all animals demonstrated deutan-type dichromatic color vision evidenced by their ability to discriminate monochromatic lights from equally-bright achromatic lights except for a “neutral point” near 505 nm. These results demonstrate that a population of functional SWS cones survived the lack of post-natal photic stimulation. However, RLR shrews differed from controls in that they were poorer at making chromatic/achromatic d~~riminations. While no severe disorgani~tion of color vision was evident, the poorer discrimination displayed by the RLR animals is likely the result of changes in post-receptoral visual mechanisms. Spectral sensitivity Color vision color channel Luminance channel

Visual development

Abnormal visual environments during early post-natal development can severely modify the normal anatomy and physiology of the mammalian visual system and impair visual function (for reviews, see Boothe, Dobson & Teller, 1985; Movshon & Van Sluyters, 1981; Sherman & Spear, 1982). One possible mechanism is that unstimulated or weakly stimulated populations of neurons are at a disadvantage in the competition for synaptic sites compared to normallystimulated cells, and consequently do not develop normally. Although research on competitive mechanisms during visual development has focused primarily on unbalanced input from the two eyes, it is likely that competitive interactions may also occur between functional channels of the visual system (e.g. W-, X- and *To whom reprint requestsshould be addressed at the Departmentof Psychology, University Louisville, KY 40292, U.S.A.

of Louisville,

Tree shrew

Tupaia

Opponent-

Y-cell channels, ON- and OFF-center channels, opponent-color and luminance channels, etc.). In humans, color vision and spectral sensitivity develop to some extent post-natally, matic opponency present by about

and trichromatic

with chrocolor vision

3 months of age (Brown, 1990; Brown & Teller, 1989; Teller & Bornstein, 1987). Since evoked potential recordings show that the short-wavelength-sensitive (SWS) cone mechanism (which appears to be the slowest to develop, Pulos, Teller & Buck, 1980), contributes to spectral sensitivity by 4-6 weeks of age (Volbrecht & Werner, 1987), it is likely that post-receptoral mechanisms of color vision are still undergoing development between 1 and 3 months. In animals, post-natal development of color vision mechanisms has also been documented in physiological studies of ground squirrels (Jacobs & Neitz, 1984; McCourt & Jacobs, 1983). Given that it has a post-natal component, the course of color vision development, and the interaction of the color-opponent and lumi-

1749

1750

HEWOOD M. PE+IXYand JOHN P.

nance channels, might be expected to be influenced by the nature of the post-natal lighting environment. Rearing animals in darkness (Boothe, Teller & Sackett, 1975; McCourt &Jacobs, 1983) or in dim white light (Di, Neitz & Jacobs, 1987) does not permanently impair the development of normal color vision and spectral sensitivity, although it can delay it. However, because darkness does not give a competitive advantage to any sub-population of visual neurons, it is not surprising that visual deficits are less pronounced in dark-reared animals in comparison to other deprivation paradigms (e.g. monocular paitem deprivation). Furthermore, although dim-light might be expected to favor the scotopic system over photopic, all classes of cone photoreceptors would be equally affected. Thus, a competitive advantage would not be afforded to either the photopic color-opponent or luminance channels. On the other hand, a spectrallybiased lighting environment that produces an unequal stimulation of the cone types (e.g. deep red light) could cause unbalanced activity in these photopic channels that in turn could promote abnormal development. Few studies have looked at the effects of spectrally-biased environments on the development of mammalian color vision. Aside from two early studies in which adult and juvenile monkeys were studied (LeGros Clark, 1943; Chow, 1955), only McCourt and Jacobs (1983) and Brenner, Schelvis and Nuboer (1985) have reared animals in red light from birth. In their physiological study of ground squirrel optic nerve fibers, McCourt and Jacobs (1983) found that significantly fewer fibers in the red-light-reared (RLR) animals received inputs from both classes of cones. This result was not seen in adults housed in red light for an equal amount of time. However, it is not known whether the population of coloropponent cells would have ultimately reached normal levels had the RLR squirrels been subsequently housed in white light, nor what type of color vision would have been present if it had been measured behaviorally. In the only behavioral study of a red-light-reared animal, Brenner et al. (1985) raised one monkey in red light from birth to 3 months of age and reported that the monkey’s ability to discriminate differently colored puddings was impaired at 5 months of age. Its color vision was not tested as an adult, although its increment-threshold spectral sensitivity (measured at about 18 months) was not different than that of its mother.

KELLY

To investigate the long-term effects of a spectrally-biased rearing environment, the present study employed traditional psychophysical methods to measure color vision and spectral sensitivity in adult tree shrews that were reared in red light. Tree shrews were chosen primarily because their highly cone-dominated retinas contain only two functional classes of cones (Jacobs & Neitz, 1986; Petry & Harosi, 1990; Polson, 1968), thus providing a simplified system for study. Based on the absorption properties of the tree shrew cone pigments (Petry & Harosi, 1990), rearing tree shrews in deep red light effectively deprives the SWS cones of photic stimulation without affecting stimulation of the other long-wave-sensitive (LWS) cone type. This differential stimulation of cone types should maximally affect the equilibrium of the tree shrew’s opponent-color channel, but should have little detrimental impact on a LWS-driven luminance channel. METHODS

Subjects Psychophysical data were collected from six tree shrews (Tupaia belangeri). All were colony reared under one of the lighting conditions described below. Poor maternal care necessitated that the three white-tight-reared shrews and one red-light-reared (RLR) animal (9-861R) be hand-reared on a liquid formula. The other two RLR shrews were mother-reared. All RLR animals received daily vitamin supplements (V&orbits, Nordeen Labs; l/4 tablet). All animals were adults when data collection was begun. White-light-rearing. Three control animals (two females and one male) were born and reared under cyclic (14 : 10, L : D) standard fluorescent lighting, (illuminance ranged from 59 to 366 lx depending upon the position of the detector in the cage). Red-light rearing. One female and two male tree shrews were born and reared in an environmentally controlled room under cyclic (12: 12, L: D) red light prod& by tungsten light sources in combination with Kodak 1A Safelightfilters (600 nm cut-off; illuminance range: 113-269 lx). Based on absorption spectra of tree shrew photopigmests (Petry & Harosi, 1990), these conditions resulted in photic stimulation of the LWS cones, and not the SWS cones or rods. A doubk door entry prevented white light stimulation when laboratory personnel entered

Color vision in red-light-reared tree shrews room. The animals were transferred as adults to the white-light colony room at which time behavioral training was begun. The period of red-light housing was 26 weeks for tree shrew 7-88-l R, 31 weeks for 9-86-l R, and 32 weeks for 6-87-2R.

the

Apparatus and optical system The animals were tested in a 12 in. square chamber using a three-alternative forced-choice paradigm. Their task was to touch one of three stimulus panels that was illuminated differently than the other two. Correct responses were reinforced with ca 0.1 ml of sweetened milk or fruit juice. The stimulus panels were mounted in a row on the front wall of the chamber which abutted a two-channel optical system. The response “keys” consisted of aluminium rectangles with a hole (0.75 in. dia) in the center of each for stimulus presentation. Stimuli were rear-projected onto glass diffusers (located immediately behind the keys), and subtended a minimum visual angle of 3.9deg (measured from the rear of the chamber). A fluorescent lamp mounted on the front wall 4 in. above the panels was used for adaptation and a tungsten lamp with diffuser above the clear ceiling provided dim ambient lighting. A laboratory microcomputer was used to randomize and time stimulus presentations, record responses and deliver reinforcements. The animal could be observed from an adjoining room through a two-way mirror. Achromatic and narrow-band chromatic lights were produced by a two-channel optical system with a 150 W xenon arc (correlated color temperature = 6000 K) as the source. Achromatic light was produced in one channef, attenuated using a variable neutral density wedge (Inconel type) and split so as to equally illuminate the 3 stimulus panels from the rear. Slight differences in the illumination of the three panels were corrected using neutral density filters (Kodak Wratten no. 96). Chromatic lights were produced in a second channel using an Instruments SA H-20 holographic grating monochromator (half-width band-pass of 8.4 nm). After passing through a variable neutral density wedge, the chromatic light was collimated and

operant

1751

directed parallel to the front wall of the chamber. Three solenoid-o~rated front-surface mirrors were positioned such that when one was raised it would simultaneously block the achromatic illumination of a stimulus panel and reflect the chromatic light in its place, the net effect being that any key could be illuminated differently than the other two (the “positive” stimulus). Solenoid-operated shutters in each channel were used to begin and end each trial, and to allow the mirrors to be raised or lowered in the absence of light stimuli. Foam cushions eliminated any noise from mirror movement and a fan that vented the xenon lamp served to mask any auditory cues. Calibrations were done using a UDT model 350 photometer~radiometer (United Detector Technology, Hawthorne, Calif.). General procedures and training Animals were initially trained to respond to any panel using an autoshaping procedure (see Domjan 8z Burkhard, 1986). Next, they were rewarded only for discriminating a brighter chromatic light from two dimmer achromatic lights. Finally, when performance was consistently above 75% correct, the intensity of the achromatic lights was adjusted to include an equal brightness value (see below) and training continued for 165 trials or until responding was above 75% correct on all keys. To enhance the reinforcement value of the liquid reward and to prevent ill-timed satiation immediately prior to training or testing, water intake was restricted for approx. 4 hr prior to beginning the session. Following the session, the animals were given ad lib water and also pieces of fruit. Neutral point test

This experiment was undertaken to determine whether the RLR tree shrews possessed color vision, and if so, to determine the location of their dichromatic neutral point. Evidence of color vision was taken to be the ability to distinguish mon~hromatic lights from achromatic lights of equal brightness. To insure conditions of equal brightness for tree shrews, achromatic lights were varied in steps of 0.1 log units* covering a range of f0.5 log units from a brightness match made by one of the experimenters. This procedure was based on Jacobs *For somewavelengthsadditional data were collected using and Neitz’ (1986) report that tree shrew brightachromatic intensity steps of 0.2 log units creating a range of’ f 1.0log units in case the kO.5 log unit range ness matches fell within 0.2-0.3 log units of might not include the brightness match. No differences matches made by human observers. The chrowere seen in the results. matic test lights were equated for brightness

1752

HEYWWDM. PETRYand JOHNP. KELLY

based on human matches to an achromatic stimulus of I56 cd/m’. Employing the method of constant stimuli, each session consisted of 55 pairings of a chromatic test light with an achromatic light of varied intensity. Five trials were run at a particular achromatic intensity, then the luminance was randomly changed to a new value for the next five trials, and so on until all 11 different intensity values were completed. The test wavelengths used were 485-500 nm in 5 nm steps, 505-520 nm in 3 nm steps, and 525-540nm in 5 nm steps. Subjects completed 1IO-220 trials per day. Data collection continued until at least six 55trial sessions were completed for each wavelen~h yielding 30 trials per wavelength/achromatic intensity combination. Spectral sensitivitytests

For this experiment, the achromatic channel was blocked so that only a single chromatic light was rear-projected on a given trial. Spectral sensitivity was measured under two adaptation andirons: (a) 1.3 lx provided by a tungsten source; and (b) a 430 lx white fluoresent lamp. &cause some of this white light was reflected on the front of the stimulus panels, the rearprojection of the chromatic light in fact resembled an increment-threshold spectral sensitivity task. The intensity of the chromatic light was varied trial-by-trial using a neutral density wedge. A modified staircase/method-oflimits ~yc~phy~~l procedure was used to maximize data collection near threshold and s~~~n~~ly insure a high level of stimulus control (Petry, Fox & Casagrande, 1984). In short, if the animal made a correct response, the intensity for the next trial was decreased by 0.1 log unit whereas an incorrect response resulted in a 0.1 log unit increase in stimulus intensity. However, two incorrect responses in a row caused an increase in intensity of 0.6 log units on the next trial. Thus the animal never went more than two trials without either receiving a reinforcement or being presented with a considerably easier task. This served to prevent “random walks” and loss of stimulus control of the animal’s responding near threshold. Data was collected in blocks of 55 trials per wavelength. The wavelength was randomly changed *Any contribution from the small populationof rods @-lo%, &tier a P&&d, 1389) is quiteunlikely and would be further minimized by the 4301x adaptation light.

from 450 to 650 nm at 10 nm intervals for each block. RESULTS

Neutral point test

Performance (percent correct) was plotted against the intensity of the achromatic light for each test wavelength studied and the lowest level of performance at each wavelength was taken to indicate that subject’s brightness match for that test wavelength. Neutral point functions were constructed by plotting the lowest percent correct values at all wavelengths. The neutral point functions obtained from two ~-lint-reared (9-86-I R and 6-8%2R) and three control tree shrews (2-86-3, 3-87-l and g-87-6) are presented in Fig. 1, and mean data for the two groups in Fig. 2. All animals demonstrated color vision by their ability to distinguish at least some chromatic lights from equally-bright achromatic lights. Performance at chance levels (33%) indicated the presence of a dichromatic neutral point near 505 nm. This is indicative of deutantype dichromacy and is consistent with previous determinations in tree shew (Jacobs & Neitz, 1986; Poison, 1968). These results show that the RLR shrews must have two functional cone systems,* however, they consistently performed more poorly throughout the spectrum, except for wavelengths beyond 525nm. To quantify these differences the midpoint and range of the neutral point were compared. Although commonly reported as a single wavelength, the neutral point is actually defined as the range of wavelengths that a dichromatic subject is able to match to white light in hue and brightness (Hsia & Graham, 1965, p, 396). In contrast to human diihromats, who exhibit a sharply defined neutral point, tree shrew neutral point functions are ‘V-shaped” (e.g. see Fig. 5.10 of Jacobs, 1981). Hence, the width of the neutral point range is directly influenced by the percent-correct criterion defined to be a “match”. If the functions are symmetrical, then the midpoint of the range (the commonly reported “neutral point”) will remain invariant regardless of the criterion chosen. However, as is evident from Fig. 1, some of the neutral point functions obtained in the present study were not symmetrical in shape. Thus, two criterion levels were used to determine the location of each animal’s neutral point: a chance-performance criterion {i.e. 33.3% correct) and thresholduerformance criterion (i.e. midwav. between the \

1753

Color vision in red-light-reared tree shrews

2-86-3

+

100 80 60 40

T

100

;

90

t; t

-s

ooPp

o-O\

b

0

20

501_ L

40

. .._-

P

20

t

3-87-1

OI

100

a0

”

60

.’

40

”

20

”

T

80

,!‘O

.’ 0

‘O-0

60..

‘0,

40

..

20

‘.

_..-_

0

0

P 9-07-6

OL: 460

Wavelength (nm)

I”

-. 490

500

:-cccccccr(

510

520

Wavelength

530

540

550

(nm)

Fig. 1. Neutral point determination for 3 control animals (open symbols) and two red-light-reared animals (solid symbols). Chance level of performance is indicated by the horizontal line at 33.3% correct.

individual animal’s maxima1 performance and chance levels). The criteria, ranges and midpoints obtained are displayed in Table 1. It is apparent from Figs 1 and 2 and from Table 1 that the RLR animals differ in several respects from control animals.* Regardless of criterion level used, the range of wavelengths that “matched” the achromatic light was broader for the RLR animals in every case. For one RLR animal (9-86-1R) the function was much shallower as well. The poorer and more variable performance of the RLR animals extended beyond the conditions of equal brightness plotted in Figs 1 and 2. This can be seen in Fig. 3, where the relative frequency of occur*That these differences represent sensory deficits and not merelybehavioral deficits is evident from the threshold criterion values in Table 1. RLR shrew 6-87-2R had the highest threshold criterion meaning that it also exhibited the highest maximum performance level. The equivalent spectral sensitivity of the two groups (see Fig. 4) also indicates that they were performing under the same level of stimulus control.

rence of percent correct scores for all of the 11 intensity values per wavelength are plotted for various wavelength ranges. Figure 3a shows performance at the neutral point where by definition, discrimination must be dependent on brightness information alone. Although both groups had difficulty with intensity discrimi-

.---.jiofZRlR 0 450

490

500

510

Wavelength

520

530

54O

@-+I 550

(nm)

Fig. 2. Mean neutral point functions for red-light-reared and control groups. Chance level of performance is indicated by the horizontal line at 33.3% correct.

1754

Hnwoo~

M. PETRYand JOHNP.

KELLY

Table 1. Neutral point ranges and midpoints Animal number Rearing condition

33% criterion” Range Midpoint

Threshold criterionb Value Midpomt Range

9-87-6 Control 2-86-3 Control 3-87-l Control Mean of controls

505.0-510.0 502.3-506.5 499.3-509.0 503.6-507.1

507.5 504.4 504.2 505.4

(60%) (60%) (55%) (58%)

497.5-513.7 496.9-510.4 495.8-517.4 496.4-513.7

505.6 503.7 506.6 505.3

9-861R Red light 687-213 Red light Mean of red light

503.7-515.3 498.8-513.0 501.3-514.2

509.5 505.9 507.7

(52%) (62%) (57%)

490.5-521.7 492.8-516.2 492.0-516.7

506. I 504.5 505.3

‘Wavelengths corresponding to the 33% correct criterion are interpolated values obtained from the individual curves shown in Fig. 1. bThe threshold criterion was defined as that percent correct score midway between chance levels (33% correct) and maximum performance of the individual animal. The wavekngth range was determined by interpolation at this criterion value.

nations at the neutral point, the poorer performance of the RLR animals demonstrates a deficit even on pure intensity discriminations.* When some wavelength information was made available for the discrimination (i.e. f 10 nm from the neutral point), control animals performed considerably better and the performance gap between the two groups widened, as seen in Fig. 3b. At the spectral extremes (> 10 nm from the neutral point) both groups performed quite well, although the RLR animals still lagged significantly. A Mann-Whitney U-test showed the performance of the two groups to be significantly different under each of these analyses b--U = 15,163, (a--U = 472, P < 0.01; P < 0.01; c-u = 11,885, P < 0.01). Spectral sensitivity tests

Frequency-of-seeing curves, based on a mean of 230 trials each, were constructed for individual animals at each wavelength by plotting the percentage of correct responses at each stimulus intensity level (subsequently referred to in log terms as the amount of neutral density added to the stimulus). In the few cases where orderly psychometric functions did not result (6 of the 60 curves plotted),.those wavelengths were omitted from that subject’s data set. Because of the psychophysical paradigm used, a set of curves was produced where the falling edge of the psychometric function (i.e. near threshold) was based on substantially more trials per point (e.g. at least 10 and as many as 45 trials per point) than the ends of the curve. A least-squares linear regression analysis was performed on the linear portion of the data, and a sensitivity value was *The equivaknt sensitivity of these RLR and control animals at the neutral point is demonstrated in the spectral set&iv&y tests, thus ruling out the possibility of skewed distributions of intensity values relative to equal brightness for the two groups.

a.5

T

0.4

0.3

0.2

0.1

0

I T Cl

0.30

RLA hl=21 Normal In=31

(b)

0.25 s “,

0.20

P

0.15

P L

0.10 0.05 0

Correct sccre t%) Fig. 3. Histograms of relative frequency of the percent correct scores obtained for all 11 intensity values per wavekngth. (a) Percent correct score dktribution at the neutral point for red-light-mamd animak (solid bars) and controls (open bars). (b) Distriitioa of scorea for wavekngtbs f 10 nm from the neutral point (i.e. 495-514 run). (c) Distribution of seotes for wavektt@s -tar than 1O~n from the neutral point (i.e. 485 and 517~53Onm). gee text for details.

Color vision in red-light-reared

*

1755

tree shrews

o--o

Z-86-3

calculated to be that density value that corre.-. s-Bb-1R 2.5 .-. 6-87-2R T sponded to a performance level of 66.7% (i.e. .-. 7-88-1R 2.0 halfway between 100% correct and chance perx c formance, 33.3%). These values were corrected .$ 1.5 'Z for relative energy of the source and the spectral c x characteristics of the optical system and plotted 1.0 m against wavelength to yield a spectral sensitivity s 0.5 curve for each subject. Spectral sensitivity functions measured under 550 800 650 500 the 16 lx adaptation condition for two red-light400 450 reared tree shrews (6-87-2R and 7-88-1R) and a control animal (2-86-3) and are shown in Fig. 4. Wavelength (nm) The individual curves are remarkably similar Fig. 5. Spectral sensitivity measured under 430 lx white light in shape, exhibiting two maxima (one at adaptation for 3 red-light-reared animals (solid symbols) and a control animal (open symbols). 450470nm and one at 550-590nm) and a pronounced minimum at 510 nm. This shape is Figure 5 shows spectral sensitivity functions characteristic of spectral sensitivity functions previously reported for tree shrews (Jacobs & for three RLR tree shrews (9-86-l R, 6-87-2R and 7-88-1R) and a control animal (2-86-3) Neitz, 1986) and human deuteranopes (Zrenner, 1983) obtained using increment-threshold tech- obtained under 430 lx fluorescent white adaptation. The functions resembled those shown in niques where the sensitivity of the colorFig. 4 with two exceptions. First, although the opponent channel (rather than the luminance channel) determines sensitivity across much of general shape of the functions was similar, there was more intra- and inter-subject variability in the visible spectrum (King-Smith & Carden, the data collected under the brighter adaptation. 1976). Since the long duration and relatively Second, there was an overall decrease in sensilarge stimulus size used in the current expertivity across the spectrum, as might be expected iment would be expected to favor the opponent color channel, it is reasonable to assume that the from the bright white adaptation light, however, curves in Fig. 4 represent the sensitivity of the the effect was differential with the short wavetree shrew’s single opponent color channel. There length lobe of the function (450-510 nm) apappeared to be a tendency for the RLR animals prox. 0.3 log units less sensitive and the long to be slightly more sensitive to longer wave- wavelength lobe (520-650 nm) about 0.5 log lengths. Mean sensitivity from 570-590 nm was units less sensitive. These differences are easily +O. 14 log units for 6-87- 1R and + 0.18 log units accounted for by the brighter adaptation light for 7-88-1R. (This difference reached as much used and by its spectral output which should as +0.42 log units for 6-87-2R at 570 nm and adapt the LWS cones more strongly than the +0.36 log units for 7-88-l R at 590 nm, but these SWS cones (standard warm white fluorescent, differences appeared to result primarily from Wyszecki & Stiles, 1967). As can be seen in opposite point-by-point fluctuations in the sensi- Fig. 5 the sensitivity of control animal 2-86-3 tivity curves of the RLR and control subjects.) is bracketed by the sensitivities of the RLR shrews, even at the trough of the function and o-0 2-86-3 at the longer wavelengths. Thus, it must be 2.5.. .-. 7-88-1R concluded that there was no absolute or relative I-. 6-87-217 >r change in spectral sensitivity as a result of e! .z 2.0.red-light rearing. Although not characteristic of .e c" I 1.5.. -.-. the RLR group in general, the broad trough shown in the function of shrew 9-86-1R is of interest considering the extremely shallow neutral point function exhibited by this animal.

I

0.00.

0.54::.::.::.:~:.::..::_::,::,

400

450

500

550

600

650

700

DISCUSSION Wavelength (nm)

Fig. 4. Spectral sensitivity measured under 16 lx white light illumination for 2 red-light-reared animals (solid symbols) and a control animal (open symbols).

The results show that color vision (based on a functional opponent-color channel mediated by LWS and SWS cones) and normal spectral

1756

HEYWOOD M. PETRYand JOHNP.

sensitivity were present in tree shrews reared for as long as 32 weeks in deep red light that prevented photic stimulation of the SWS cones. The performance of these animals on chromatic/achromatic discriminations, however, was considerably poorer than that of normals. In light of the severe visual deficits that can be produced by other types of differential visual stimulation during development, it is interesting that the RLR animals performed as well as they did considering the prolonged lack of photic stimulation of the SWS cones. One reason for this difference is probably activity-dependent. Because color-opponent cells by definition receive input from more than one cone mechanism, input from the nondeprived LWS cones would insure post-natal activity of these cells. It is extremely unlikely that the critical period for color vision development in tree shrews extends beyond the period of red-light-rearing. Tree shrews are sexually mature by 26 weeks of age and the 26-32-week red-light-rearing period well encompassed the &lZweek period during which post-natal development of color vision is thought to occur in primates (Boothe et al., 1975). Also, monocular pattern deprivation produces visual deficits in tree shrews within the first 18 weeks of life (Casagrande, Guillery & Harting, 1978; Norton, Casagrande & Sherman, 1977). The fact that the spectral sensitivity of the RLR shrews was equivalent to that of control animals on both an absolute and relative scale shows that red-light-rearing did not cause a permanent imbalance of cone inputs to the opponent-color channel. King-Smith and Carden (1976) have shown that the detection of spectral lights is mediated through probability summation by the most sensitive of the parallel luminance and color-opponent channels. The stimulus parameters employed in the present experiments favored the color-opponent channel and the fact that the shapes of the spectral sensitivity curves indicate a subtractive interaction of the SWS and LWS cones, clearly demonstrates detection by the color-opponent mechanism. Any significant variation in the relative weights of the SWS and LWS cone inputs to this channel in RLR animals compared to normals should have been apparent in the relative heights of the short-wave and longwave peaks. Aside from a tendency for the RLR shrews to be slightly more sensitive from 570-590, no differences were seen in the shapes

KELLY

of the curves for the two groups. These results are in accordance with Brenner et al.‘s (1985) report that increment spectral sensitivity was normal in a monkey reared from birth to 3 months in red light. Together, these results suggest that the mechanisms of increment spectral sensitivity are likely to be “hard-wired” and not subject to the influence of the spectral lighting environment during post-natal development. Despite apparently normal spectral sensitivity and the exhibition of dichromatic color vision, the RLR shrews were clearly poorer at making chromatic/achromatic discriminations. The exact mechanism for this de&it is unknown, but is likely to involve higher level interactions of the opponent-color and luminance channels. Changes in the equilibrium of post-receptoral opponent-color mechanisms have been demonstrated psychophysically in humans when appropriate chromatic adaptation paradigms were used (Jameson, Hurvich & Vamer, 1979; McCollough, 1965). These and other studies have shown that even temporary adaptation can produce sizable and long-lasting shifts in the equilibrium of post-receptoral opponent color mechanisms. in the RLR tree shrews, a repeated biasing of post-receptoral opponent-color cells towards red may have resulted in the formation of fewer synaptic connections with other coloropponent and luminance signalling neurons or may have caused the establishment of abnormal connections. Evidence for environmental influence on synapse formation has been reported by Mackay and Bedi (1987) who found decreased synapse-per-neuron ratios for cells in the superior colliculus of rats “dark-reared” in very dim red light, and by Schoups and Black (1990) at the molecular level. If indeed this is the case, such changes could have affected the RLR animals’ color perception, which is unfortunately difficult to assess in nonverbal subjects. In addition, the poorer intensity discrimination by the RLR tree shrews at their neutral point (where the color-opponent channel is at equilibrium) is indicative of a deficit in the luminance signalling channel. Taken together, these results suggest that although the mechanisms of color opponency and spectral sensitivity in general were relatively the unaffected by the red-light-rearing, neural mechanisms which mediate chromatic/ achromatic discriminations were subject to influence by the spectral environment during development.

Color vision in red-light-reared Acknowledgements-This work was supported by NIH grant EY-07113 to HMP and by BRSG funds obtained through SUNY at Stony Brook. The authors are grateful to K. Bjorn, D. Dougherty, M. Grayson, I. Perretti and C. Williams for data collection, C. Lief and D. Morais for assistance with red light rearing, G. Hudson and R. Reeder for technical assistance, and Drs K. Knoblauch and J. G. May III for helpful discussions. Some of these data were presented in different form at the I989 annual meeting of the Association for Research in Vision and Ophthalmology (Kelly & Petry, 1989).

tree shrews

1151

Kelly, 3. P. & Petry, H. M. (1989). The effects of red-lightrearing on cotor vision in the tree shrew. ~~est~gative Ophthalmology and Visual Science (Suppl.), 30, 126, King-Smith, P. E. & Carden, D. (1976). Luminance and opponent color contributions to visual detection and adaptation and to temporal and spatial integration. Journal of the Optical Society of America, 66, 709-717.

LeGros Clark, W. E. (1943). The anatomy of cortical vision. Transactions of the Ophthalmological Society, 24, 229-245.

Mackay, D, & Bedi, K. S. (1987). The combined effects of unilateral enucleation and rearing in a “dim” red light on synapse-to-neuron ratios in the rat superior colliculus. Journal of Comparafive Neurology, 256, 444-453.

REFERENCE Boothe, R., Teller, D. Y. & Sackett, G. P. (1975). Trichromacy in normally reared and light deprived infant monkeys (~acacu nemesrrina). Vision Research, 15, 1187-1191. Boothe, R., Dobson, V. & Teller, D. Y. (1985). Postnatal development in vision in human and nonhuman primates. Annual Review o~~~roscienee,

8, 495-545.

Brenner, E., Schelvis, J. & Nuboer, J. F. W. (1985). Early coiour deprivation in a monkey (Macaca ~~cicu~ar~). Vision Research, 25, 1337-1339. Brown, A. M. (1990). Development of visual sensitivity to light and color vision in human infants: A critical review. Vision Research, 30, I1 59-I 188. Brown, A. M. &Teller, D. Y. (1989). Chromatic opponency in 3-month-old human infants. Vision Research, 29, 37-45.

Casagrande, V. A., Guillery, R. & Harting, J. K. (1978). Differential effects of monocular deprivation seen in different layers of the lateral geniculate nucleus. Jouma! of Comparative Neurology, 179, 469-486.

Chow, K. L. (1955). Failure to demonstrate changes in the visual system of monkey kept in darkness or in colored lights. Journaf of Comparative Neurology, 102, 597-606. Di, S., Neitz, J. & Jacobs,

G. H. (1987). Early color deprivation and subsequent color vision in a dichromatic monkey. Vision Research, 27, 2009-2013. Domjan, M. & Burkhard, B. (1986). The principles of fearning and behavior. Monterey, Calif.: Brooks/Cole. Hsia. Y. & Graham, C. H. (1965). Color blindness. In Graham, C. H. (Ed.), Vision and visual perception (pp. 395-413). New York: Wiley. Jacobs, G. H. (1981). Comparative color vision. New York: Academic Press. Jacobs, G. H. & Neitz, J. (1984). Development of spectral mechanisms in the ground squirrel retina following lid opening. Experimental Brain Research, 55, 507-514. Jacobs, G. H. & Neitz, J. (1986). Spectral m~hanisms and color vision in the three shrew (Tupaia belangeri). Vision Research, 26, 291-298.

Jameson, D., Hurvich, L. M. & Varner, F. D. (1979). Receptoral and postreceptoral visual processes in recovery from chromatic adaptation. Proceedings of the National Academy of Sciences,

U.S.A.,

76, 3034-3038.

McCollough, C. (1965). Color adaptation of edge detectors in the human visual system. Science, 149, 1115-l 116. McCourt, M. E. & Jacobs, G. H. (1983). Effects of photic environment on the development of spectral response properties of optic nerve fibers in the ground squirrel. Experimental Brain Research, 49, 4434.52.

Movshon, J. A. & Van Sluyters, R. C. (1981). Visual neural development. Annual Review of Psychology, 32,417-522. Muller, B. L Peichl, L. (1989). Topography of cones and rods in the tree shrew retina. Journal of Comparative Neurology, 282, 58 I-594.

Norton, T. T., Casagtande, V. A. & Sherman, S. M. (1977). Loss of Y-cells in the lateral geniculate nucleus of monocularly deprived tree shrews. Science, 197, 786786. Petry, H. M. & Harosi, F. I. (1990). The visual pigments of tree shrew (Tupaiu belangeri) and galago (Galago crassicaudatus): A microspectrophotometric investigation. Vision Research, 30, 839-85 1. Petry, H. M., Fox, R. % Casagrande, V. A. (1984). Spatial contrast sensitivity of the tree shrew. Vision Research, 24, 1037-1042.

Poison, M. C. (1968). Spectral sensitivity and color vision in Tupaia glis. Doctoral dissertation, Indiana University, Bloomington. Pulos, E., Teller, D. Y. & Buck, S. L. (1980). Infant color vision: A search for short wavelength sensitive mechanisms by means of chromatic adaptation. Vision Research, 20, 485-493.

Schoups, A. A. & Black, I. B. (1990). Environmental lighting regulates development of synaptic molecules in rat visual cortex. Investigative Ophthalmology and Visual Science (Suppl.), 31, 58.

Sherman, S. M. & Spear, P. D. (1982). Organization of visual pathways in normal and visually deprived cats. Physiological Reviews, 62, 738-855.

Teller, D. Y. 8r Bornstein, M. H. (1987). Infant color vision and color perception. In Handbook of infant perception (Vol. 1, pp. 185-236). Orlando, Fla: Academic Press. Volbrecht, V. J. % Werner, J. S. (1987). Isolation of short-wavelength-s~sitive cone photot~ptors in 4-6week-old human infants. Vision Research, 27, 469-478. Wyszecki, G. & Stiles, W. S. (1967). Color science: Concepts and methods, quantitative data and formulas. New York: Wiley. Zrenner. E. (1983). Neurophysiological aspects ofcolor vision in primates. Berlin: Springer.