Pattern and Motion Vision in Cats with Selective Loss of Cortical Directional Selectivity

Transcription

1 The Journal of Neuroscence Aprl 1986, 6(4): Pattern and Moton Vson n Cats wth Selectve Loss of Cortcal Drectonal Selectvty Tatana Pasternak and Laura J. Lenen Center for Vsual Scence and Center for Bran Research, Unversty of Rochester, Rochester, New York Neurons n the vsual cortex of cats reared n 8 Hz stroboscopc llumnaton show a profound loss of drectonal selectvty, but no detectable defcts n orentaton selectvty, contrast senstvty, and temporal frequency response, and only a slght reducton n spatal resoluton. In the present study, spatal vson, temporal resoluton, and a varety of moton detecton and dscrmnaton thresholds were examned behavorally n such cats. These psychophyscal measurements revealed nearly normal spatal and temporal vson, but severe abnormaltes n vsual dscrmnatons based on dfferences n stmulus drecton. Specfcally, strobe-reared cats showed normal orentaton dscrmnaton and temporal frequency resoluton, nearly normal contrast senstvty at low spatal frequences, and a slght reducton of senstvty to hgh spatal frequences. At hgh contrasts, the cats were able to dscrmnate opposte drectons of moton over a wde range of vsble speeds, and ther performance was ndstngushable from that of normal cats. However, a comparson of contrast thresholds for detectng movng gratngs and for dscrmnatng ther drecton of moton revealed severe abnormaltes n strobe-reared anmals. At low spatal frequences (.28 cycles/deg), normal cats could dscrmnate the drecton of gratng moton at contrasts that were just barely vsble, whereas the strobe-reared cats could detect the gratng at contrasts smlar to those requred by normal cats, but requred contrasts about 1x the threshold to dentfy the drecton of moton. Normal cats showed nearly dentcal contrast senstvty for detectng and dscrmnatng gratngs of hgh spatal frequency at hgh temporal frequency (drft rates), but when the temporal frequency was low, ther senstvty for detecton exceeded that for drecton dscrmnaton. Wth the same stmulus parameters, the detecton senstvty of strobe-reared cats was smlar to that of normal cats, but ther senstvty for drecton dscrmnaton was further reduced. These results provde the frst evdence of an nvolvement of drectonally selectve neurons n drecton dscrmnaton. Conversely, they show that even a small number of drectonally selectve neurons can support normal drecton dscrmnatons f the targets are of hgh contrast. Many cells n the cat vsual cortex are drectonally selectve,.e., they respond strongly to patterns movng n one drecton, but weakly or not at all to the same pattern movng n the opposte drecton (e.g., Hubel and Wesel, 1962; Orban et al., 198 la; Spear and Baumann, 1975). It s commonly assumed that such drectonally selectve neurons are the physologcal substrate of psychophyscally dentfed drectonally selectve Receved May 6, 1985; revsed Aug. 19, 1985; accepted Aug. 21, Ths research was supported by NE1 Grants EY 4118 to T.P. and P3 EY to the Center for Vsual Scence. We thank Dave Kurtz, Deborah Gallant and Jeff Lewne for excellent techncal assstance and Walter Makous, Bll Me&an and Gary Sclar for comments on the manuscrpt. Correspondence should be. addressed to Tatana Pastemak at the above address. Copyrght 1986 Socety for Neuroscence /86/4938-8$2./O mechansms (e.g., Sekuler et al., 1978; Wlson, 1985). However, there has been no drect demonstraton of the nvolvement of these neurons n vsual percepton. Some years ago, Cynader and Chemenko (1976) reared cats n hgh-frequency (8 Hz) stroboscopc llumnaton for a perod of 6 to 8 months and reported a dramatc loss of drectonal selectvty n cortcal cells that otherwse showed normal orentaton selectvty and other receptve feld propertes. Pasternak et al. (1985a) subsequently confrmed and extended ths fndng. They gathered data from area 17 of cats reared n 8 Hz stroboscopc llumnaton and found more than a 9% loss n the number of drectonally selectve cells, but ntact orentaton selectvty, contrast senstvty, and temporal frequency response and only a slght reducton n the number of cells respondng to hgh spatal frequences. Thus, cats reared n 8 Hz stroboscopc llumnaton provde a useful anmal model for explorng vsual performance n the vrtual absence of cortcal drectonal selectvty, but wth otherwse largely ntact cortcal receptve feld propertes. The present study was desgned to examne the wdely accepted noton that drectonally selectve cortcal cells consttute the neural bass of moton percepton. We used behavoral technques to examne the spatal and temporal vson of such cats, as well as ther ablty to make dscrmnatons based on stmulus drecton. The results of sngle-unt recordngs from area 17 and the lateral suprasylvan area n these anmals have been reported elsewhere (Pastemak et al., 1985a; Spear et al., 1985). Some of the present behavoral results have been brefly reported elsewhere (Pastemak and Lenen, 1984; Pastemak et al., 1984, 1985a). Materals and Methods Subjects Twelve cats of both sexes were reared n a room llumnated at a frequency of 8 Hz by a 3 rsec stroboscopc flash. The cats were exposed to ths llumnaton for 12 hr each day and were otherwse n total darkness. They were placed n ths envronment durng the frst week of lfe, before eye openng, and remaned there for eght months before beng removed to a normally llumnated room shortly before behavoral testng. Durng testng, the cats were mantaned at 845% of ther normal body weght. Water was contnuously avalable n ther home cage, and they receved a daly food supplement of Purna Cat Chow. Followng the completon of behavoral testng, sx of the cats were used for sngle-unt physologcal studes, the results of whch are reported elsewhere (Pastemak et al., 1985a; Spear et al., 1985). General behavoral procedure A two-alternatve, forced-choce procedure was used n all experments. The cats were placed n a test chamber equpped wth two glass response panels located sde by sde above the feeder. Stmul were vewed through the response panels. At the begnnng of each tral, the cats were presented wth two stmul, each located behnd one of the response panels. After a 2 set vewng perod, the frst nose-press response towards the 938

2 The Journal of Neuroscence Vson Wthout Cortcal Drectonal Selectvty 939 correct stmulus was rewarded wth a small amount of pureed beef. An ncorrect response resulted n a 1 set perod, sgnaled by a loud tone, durng whch no reward was delvered and responses were neffectve. The ntertral nterval ranged from 5 to 8 set n dfferent experments. More than three consecutve errors on the same sde evoked a correcton procedure: The stmulus appeared on the nonpreferred sde untl the anmal made an approprate response. Correcton trals were not ncluded n the data analyss. Each sesson conssted of 2 trals. In each experment, the cats were traned on a smple dscrmnaton task untl performance of over 8% correct was acheved n four consecutve sessons, or 9% correct or better n three consecutve sessons. After ths crteron was met, threshold measurements were ntated. Threshold measurements In all experments, thresholds were measured wth a starcase procedure: Each correct response produced a decrease n stmulus contrast (or orentaton dfference or stmulus speed) wth a probablty of.3, and each ncorrect response produced an ncrease n stmulus contrast (or orentaton dfference or stmulus speed). Ths assured mantenance of the anmal s overall performance for the sesson at approxmately 75% correct. Thresholds were taken from the resultng psychometrc functons at a stmulus value correspondng to 75% correct. Contrast senstvty for statonary gratngs These methods are smlar to those descrbed prevously (Pastemak and Mergan, 1981; Pastemak et al., 1983). Brefly, a statonary, vertcal snusodal gratng and a comparson unform feld of 17 cd/m2 were generated on two Tektronx 66 osclloscopes (P-3 1 phosphor). Gratng onset was modulated bv a half-cvcle of a rased cosne of 1.25 Hz. Each stmulus subtended an 11 x l$ vsual angle and had an equal-lumnance whte surround extendng outward 22. Contrast was defned as L ma - L,,,~Lmax + -L, where Lx s the lumnance of the brght bar of the gratng and L,,, s the lumnance of the dark bar. Durng prelmnary tranng, the cats were traned to dscrmnate between a hghcontrast (.8) low spatal frequency (.33 cycles/deg) gratng and a blank feld of the same mean lumnance. After the cats mastered ths task, vsual acuty and then contrast thresholds were measured. Acuty was measured by varyng the spatal frequency of gratngs at.8 contrast, whle contrast thresholds were measured by varyng contrast at the followng spatal frequences:.2,.26,.33,.44,.55,.77,.98, 1.3, and 1.6 cycles/deg. Contrast senstvty (recprocal of contrast threshold) was measured n eght strobe-reared and fve normal cats. For each anmal the sequence of testng at each spatal frequency was rregular, and testng contnued untl no further mprovement n thresholds was seen. At least four thresholds were determned at each frequency. Crtcal-flcker frequency (CFF) The detals of the procedure have been descrbed elsewhere (Pastemak et al., 1985b). CFF was measured wth the same dsplay system and procedures as those descrbed above for acuty testng. The cats vewed two unpatterned felds of the same mean lumnance (17 cd/m2), one of whch was snusodally modulated n tme. They were traned to respond to the flckerng feld. Durng ntal tranng, the cats dscrmnated between a 7 Hz stmulus and a nonflckerng feld. Thresholds were measured by varyng temporal frequences at the depth of modulaton of.8 (depth of modulaton s the temporal equvalent of contrast defned above). Orentaton dscrmnaton The stmul were two low-frequency (.32 cycles/deg), hgh-contrast (.8) statonary gratngs projected onto crcular felds subtendng 2 of vsual angle. The mean lumnance of gratng patterns was 3.8 cd/m*. The orentaton ofeach gratng was controlled by means of a dove prsm. Durng ntal tranng the cats vewed a vertcal (9 ) and horzontal (OO) gratng, and they were rewarded for choosng the vertcal. Once the cats mastered the task, threshold measurements began. The orentaton of the horzontal gratng was vared n equal steps of.2 log unts between horzontal and near-vertcal (81., 57.6,39.6,28.8, 19.8, 14.4, 1.8, 7.2, 5.4, and 3.6 deg) n a starcase procedure; daly orentaton dfference thresholds were determned from the resultng psychometrc functons. Three normal and two strobe-reared cats were tested n ths task. IOOr I Spatal Frequency k/d) Fgure 1. Contrast senstvty of fve normal (open symbols) and eght strobe-reared (flled symbols) cats. Each pont s a geometrc mean of contrast senstvtes of each cat. The error bars are SEM. SEMs for ndvdual anmals were, on the average,.4 for normal cats and.6 for strobe-reared cats. Data ponts at the hghest spatal frequency (trangles) were determned wth gratngs of.8 contrast. Mean lumnance, 17 cd/ml. Moton dscrmnatons at hgh stmulus contrasts The technques of stmulus generaton and other methodologcal detals have been orevouslv descrbed (Pastemak and Meraan. 198, 1984). Low-speed^thresholds for the detecton of moton and drecton dscrmnaton were tested wth hgh-contrast (.8), sotropc random dot patterns and square-wave gratngs back-projected onto two 2 crcular screens. The mean lumnance of dot patterns was.95 cd/m2, whle that of gratngs was 3.8 cd/m2. In the moton detecton task, the cats vewed two random dot felds, one of whch moved contnuously whle the other was statonary. All cats were rewarded for respondng to the movng stmulus wth the excepton of two cats (81 and 87) whch were rewarded for choosng the statonary one. For cats 82,85, 83, and 8 11, the movng stmulus drfted to the left; for the remanng cats, the stmulus drfted to the rght. Durng ntal tranng, the movng stmulus was at a sngle, relatvely hgh (25 deg/sec) speed. Measurements of lowspeed thresholds were ntated after the cats relably dscrmnated between the movng and the statonary patterns. As n the other tests, stmulus speed was vared n a starcase procedure and thresholds were taken at 7 5% correct. In the drecton dscrmnaton task, the cats vewed two patterns movng at the same speed, but n opposte drectons (rght and left). The cats were traned to respond towards the rghtward stmulus. Low-speed threshold measurements for ths dscrmnaton began after the cats had mastered the task. In one normal (8) and one strobereared (84) cat, low-speed thresholds for drecton dscrmnaton were also measured wth hgh-contrast (.8) square-wave gratngs at three spatal frequences:.32,.58, and 1.1 cyclesldeg. Contrast senstvty for detecton and drecton dscrmnaton Snusodal gratngs were generated on a Hewlett-Packard 1332A osclloscope screen (P-3 1 phosphor) placed behnd two 12 -dameter crcular openngs n the whte surround. The mean lumnance of the stmul and the surround was 75 cd/m2. In the detecton experment, the cats vewed a movng snusodal gratng and a blank feld of the same mean lumnance. They were rewarded for respondng toward the feld contanng the gratng. In the dscrmnaton task, the cats vewed two dentcal snusodal gratngs movng horzontally n opposte drectons. Low spatal frequency gratngs (.28 cycles/deg) movng at 8 deg/sec (2.2 Hz) were used durng ntal tranng. The cats were rewarded for choosng the feld contanng the rghtward gratng. In both tasks, contrast senstvty was measured over a range of spatal frequences and stmulus speeds. All cats were tested at.28,.46, and.77 cycles/deg at the follownn sneeds: and n some cases. 1 or 64Vsec. In addton, one of the normal cats (cat 113) was tested at 1.3 cyclesldeg, and a second normal cat (116) at.61 cyclesjdeg. For each cat, both detecton and dscrmnaton thresholds were measured on the same day, and n many cases the same spatal and temporal frequences were used.

3 94 Pasternak and Lenen Vol. 6, No. 4, Apr Fgure 2. Moton detecton and drecton dscrmnatons at hgh stmulus contrasts (.8). A (top left), Dscrmnaton of opposte drectons of moton by one of the strobe-reared cats (84). The data for all stmulus speeds were collected n a sngle, 2~tral sesson. B (bottom), Low-speed thresholds for detecton (open columns) and drecton dscrmnaton (strped columns) for normal (ZeJ> and strobe-reared (rght) cats. Error bars are SEM. Stmul were hgh-contrast (.8) random dot patterns. C (top rght), Low-speed thresholds for normal (open symbols; cat 8) and a strobe-reared (tlled symbols; cat 84) subject as a functon of spatal frequency. Stmul were hgh-contrast (.8), square-wave gratngs. A. 6 1 l 2 l l 5 Cl Q 6 4. I 1 I STIMULUS SPEED (deg/sec) NORMALS CAT 8 CAT 73 t t l SPATIAL FREQUENCY (c/deg) STROBE-REARED CAT 82 CAT 84 CAT 81 Results Spatal vson Contrast senstvty for statonary gratngs The mean contrast senstvty (crcles) and acuty (trangles) of fve normal and eght strobe-reared cats s shown n Fgure 1. Although the senstvty of the two groups s smlar at lower spatal frequences, that of strobe-reared cats s reduced at medum and hgh spatal frequences. Dfferences n senstvty between the two groups range from.2 to.5 log unts. Acuty estmates were also hgher for normal (mean 2.53 cycles/deg) than for strobe-reared cats (mean 1.98 cycles/deg). It s worth notng that, for both groups, acuty estmates obtaned by varyng spatal frequency of gratngs at a constant contrast of.8 are consstent wth contrast senstvty estmates obtaned by varyng gratng contrast at a gven spatal frequency. Although the majorty of the deprved cats were less senstve to hgh spatal frequences than the normals, there were two anmals (cats 8 11 and 8 12) wth acuty and/or senstvty overlappng that of normal cats. Table 1. Orentaton dfference thresholds and crtcal flcker frequency (CFF) n normal and strobe-reared cats Orentaton threshold (da) Normal cats 26.6 (.77) 24.4 (.93) 17.5 (.84) Strobe-reared cats 24.8 (2.16) 16. (.83) a Numbers n parentheses are SEM. CFF (Hz) 38.3 (.83) 42.4 (1.1) 43.2 (1.1) 36. (.63) 44.1 (.67) Orentaton deference thresholds We measured orentaton dfference thresholds for.32 cycles/ deg gratngs for three normal (cats 7, 14, and 73) and two strobereared cats (87 and 81). Mean orentaton thresholds for each cat are shown n Table 1. There are no obvous dfferences between the two groups; orentaton thresholds for all anmals range from 16 to These results show that rearng cats n 8 Hz stroboscopc llumnaton does not dmnsh ther ablty to dscrmnate dfferences n stmulus orentaton. Temporal resoluton Crtcal flcker frequency (CFF) was tested n three normal cats (19, 111, and 128) and two strobe-reared cats (cats 8 11 and 8 12); the thresholds are shown n Table 1. The flcker resoluton of the two strobe-reared cats was nearly dentcal to that found n normal cats, and ranged between 36 and 44 Hz. Thus, we found no detectable abnormaltes n temporal resoluton n strobe-reared cats. Moton thresholds Detecton and dscrmnaton at hgh stmulus contrasts Fgure 2A shows the drecton dscrmnaton performance of one of the strobe-reared cats wth respect to hgh-contrast random dot patterns. The cat dscrmnated opposte drectons of moton at 9% correct or better over a wde range of speeds (1 O -8 /sec). At lower speeds, ths anmal and the other strobereared cats also performed the task as well as the normal cats; and even at the slowest detectable speed (low-speed thresholds), they were able to correctly dentfy the drecton of movng random dot patterns. Ths s shown n Fgure 2B. Low-speed thresholds for the two normal (left) and three strobe-reared (rght) cats for the detecton of moton (open bars) and for drecton dscrmnaton (strped bars) are qute smlar. Thus, normal and deprved cats show smlar low-speed thresholds (lo-y/set),

4 The Journal of Neuroscence Vson Wthout Cortcal Drectonal Selectvty 941 and both groups can dentfy the drecton of moton at ths threshold speed. Random dot patterns are aperodc stmul that contan clusters of randomly dstrbuted dots. The postons of local features n these movng patterns are not constant, and change from moment to moment. Thus, the two comparson stmul dffer not only n moton drecton but also n spatal phase. To test the possblty that the cats used these phase dfferences to extract drectonal nformaton, we measured moton thresholds wth perodc gratng patterns. Although phase dfferences between the two movng gratngs may stll be present (snce the two comparson gratngs are not phase-locked), the extracton of ths phase cue may be much more dffcult, partcularly at hgher spatal frequences (Burr, 198). The lowest speeds at whch a normal and a deprved cat correctly dentfed the drecton of movng gratngs are shown n Fgure 2C. The thresholds for the normal and the strobe-reared cat are smlar at all spatal frequences and are close to the values obtaned wth random dot patterns for these cats. Even at a relatvely hgh spatal frequency (1.1 cycles/deg) the strobe-reared cat was able to dscrmnate dfferences n drecton. Thus, dfferences n spatal phase between the two comparson random dot patterns are unlkely to account for the ablty of strobe-reared cats to dscrmnate opposte drectons of moton. Drecton dscrmnaton at threshold contrast The ablty of strobe-reared cats to dscrmnate opposte drectons of moton s surprsng, gven ther profound loss of drectonally selectve neurons. However, t s possble that a few such neurons suffce to medate drecton dscrmnaton of hgh-contrast patterns. At stmulus contrasts near the detecton threshold, though, the small number of drectonally selectve cells stll present n strobe-reared cats may not be suffcent, and drecton dscrmnaton may be mpared under such condtons. Therefore, we measured contrast thresholds for detecton of movng gratngs and for dscrmnaton of ther drecton of moton. Examples of psychometrc functons obtaned n the detecton/dscrmnaton experment for one normal (113) and one strobe-reared (81) cat are shown n Fgure 3. The data for each functon were collected n a sngle sesson for a.28 cycles/ deg gratng movng at 4.48 Hz (16 Isec). The detecton functons for the two cats are smlar, n terms of both ther slope and ther locaton on the contrast axs. On the other hand, the psychometrc functons for drecton dscrmnaton for the two cats dffer greatly; the functon for the deprved cat s shfted toward hgher contrasts and has a shallower slope. The deprved cat requred a 1 x hgher contrast to dscrmnate the drecton of the movng gratng than dd the normal cat. Contrast senstvtes for the two groups of cats were determned from such psychometrc functons (contrast thresholds taken at 75% correct performance) obtaned at several spatal and temporal frequences. The results of such measurements on two normal cats are shown n Fgure 4. Both cats show nearly equal contrast senstvty for the two types of thresholds at the lowest spatal frequency (.28 cycles/deg) at all temporal frequences (drft rates). At hgher spatal frequences, senstvty for dscrmnaton decreased relatve to senstvty for detecton, partcularly at the lowest temporal frequences. At.46 cycles/deg, cat 116 showed a slght decrease n senstvty at 1.84 Hz (4 /sec). Ths dfference between detecton and dscrmnaton ncreased at.61 cycles/deg, and fnally, at.77 cycles/deg, senstvty for detecton exceeded that for dscrmnaton at all speeds. The pattern of results for cat 113 was smlar, although dfferences between detecton and dscrmnaton emerged at hgher spatal frequences than for cat 116. The results for ths cat at 1.3 cycles/ deg are smlar to the data obtaned at.61 cycles/deg for cat 4J t. r go- 8 = 6- &J k 7-2 : 6-5 -? DETECTION I * 1.I.Ol.OOl Contrast Fgure 3. Examples of psychometrc functons for detecton (top) and drecton dscrmnaton (bottom) for normal (open symbols, eat 113) and strobe-reared flled symbols, cat 8 1) subjects. Each functon was determned n a sngle testng sesson for a.28 cycles/deg snusodal gratng movng at 4.5 Hz (16Vsec). The data were ftted wth a Webull dstrbuton fu&ton (Qudk, 1974) by an teratve maxmum-lkelhood estmaton procedure (Watson, 1979). The functons were obtaned wth a starcase procedure; hence, the number of trals at each contrast s not dentcal, wth the smallest number of trals at the hghest and the lowest stmulus contrast At these spatal frequences, senstvtes for dscrmnaton and detecton were dentcal only at the hghest stmulus speeds. The data for the three strobe-reared cats are shown n Fgure 5. Peak contrast senstvty for detecton (open crcles) for the three anmals ranged from 6 n the least senstve cat (84) to about 14 n the other two anmals. At.28 cycles/deg, senstvtes of the deprved and normal cats overlapped at the lowest drft rates, but at hgher rates, senstvtes of normal cats appeared somewhat hgher than those of the three strobe-reared cats. Ths dfference was not apparent at hgher spatal frequences (e.g.,.77 cycles/deg), and the functons for the two normal cats and the two most senstve strobe-reared cats are nearly dentcal. On the other hand, senstvty for drecton dscrmnaton (flled symbols) was greatly reduced n the three deprved cats. At.28 cycles/deg contrast, senstvty for dscrmnaton was almost 1 x lower than that for detecton. At a hgher spatal frequency (.46 cycles/deg), senstvty for drecton was even more reduced; and at the lowest drft rates, the performance of cats 84 and 81 dropped below 75% correct even at the hghest contrast used (.66). Fnally, at.77 cycles/deg, the cats were qute unable to dentfy the drecton of moton, even though they could detect the gratngs as well as dd normal cats. Fgure 6 llustrates the dfferences between performance on detecton and drecton dscrmnaton tasks for both normal and strobe-reared cats as the rato of contrast senstvty for dscrmnaton to detecton. Mean ratos for one normal (open symbols) and three strobe-reared (flled symbols) cats are plotted

5 942 Pasternak and Lenen Vol. 6, No. 4, Apr Cat 113 Cat 116 loo- Q.28 C/d I I.46 C/d ee.oq %.48 C/d.77 c/e I.3 c/c.77 C/N. 1 l 1 Q o odo!i!! Drft Rate (Hz) b Oo. t I. I.. I Drft Rate (Hz1 Fgure 4. Contrast senstvty as a functon of drft rate for detecton (open symbols) and drecton dscrmnaton tlled symbols) for the two normal cats. For each cat, the data were obtaned at four spatal frequences (upper rght corners). Cat 116 could not dscrmnate drectons at.6 1 cycles/deg at the lowest drft rate; hence, contrast thresholds could not be determned. Error bars (SEM) are approxmately the sze of the symbols. Mean lumnance, 75 cd/m2. as a functon of temporal frequency (drft rate) at several spatal frequences (ndcated by numbers next to each functon). In the normal cat, the rato depends on the temporal and spatal frequency of the gratng; at low spatal and hgh temporal frequences (drft rates), the rato s near 1.O whereas at the hghest spatal frequency and the slowest drft rates the rato drops dramatcally. In strobe-reared cats, the rato never exceeded.2 and vared lttle wth spatal or temporal stmulus parameters. Dscusson The behavoral defcts of strobe-reared cats are consstent wth the results of sngle unt recordngs (Cynader and Chemenko, 1976; Pasternak et al., 1985a; Spear et al., 1985) n beng restrcted largely to dscrmnatons of opposte drectons and the detecton of hgh spatal frequences. Strobe-reared cats had normal contrast senstvty at low spatal frequences, but a slght loss of vsual acuty and senstvty to hgher frequences. Thresholds for orentaton dscrmnaton, temporal frequency resoluton, and moton detecton and opposte-drecton dscrmnatons, measured wth hgh-contrast patterns, were essentally ntact; however, measurements of threshold contrast for dscrmnatng drecton revealed profound defcts. Furthermore, strobe-reared cats were able to dscrmnate drectons only at lower spatal frequences and only at contrasts that exceeded the detecton threshold by approxmately one log unt. Pattern vson Contrast senstvty The loss of hgh-frequency senstvty and acuty n strobe-reared cats s consstent wth prevous fndngs of a reducton n the proporton of observed neurons n area 17 that responded to hgh spatal frequences (Pastemak et al., 1985a). In those studes, there was a shft of less than one octave toward lower preferred frequences and a loss of spatal frequency resoluton. The behavorally measured loss n acuty (.4 octave) and the shft n peak senstvty (.8 octave) observed behavorally were of smlar magntude. Ths hgh-frequency loss s not surprsng, snce the llumnaton of the rearng envronment s sgnfcantly lower than that used for normally reared anmals, and the vsblty of hgh spatal frequences s reduced under low lumnance condtons n both humans and cats (see Pastemak and Mergan, 1981). In addton, the short flash duraton (3 +ec) probably further reduced the vsblty of spatal patterns (Arend, 1976; Nachmas, 1967; Tulunay-Keesey and Jones, 1976). Indeed, deprvaton of stmulaton by hgh spatal frequences reduces spatal resoluton. When cats are reared wth optcal blur, a reduced number of cells respond to hgh spatal frequences (Eggers and Blakemore, 1978). Behavoral studes of cats and monkeys rased wth monocular blur nduced ether by atropne or hgh power negatve lenses have also shown a loss of senstvty of the deprved eye to hgh spatal frequences (Boothe et al., 1983; Graves and Morse, 1982; Smth et al., 1983). In order to assess the stmulaton of strobe-reared cats by hgh spatal frequences n 8 Hz stroboscopc llumnaton, we estmated the gratng acuty of three human observers under condtons smlar to those experenced by cats durng rearng. Acuty for a hgh-contrast (.8), square-wave gratng, llumnated exclusvely by a 3 Fsec stroboscopc flash every 125 msec (8 Hz), was reduced by approxmately.7 to.9 octave relatve to acuty for gratngs measured n a room smlar to that n whch the normal cats were housed. Ths result was compared to the acuty/lumnance functon of a human observer, suggestng an equvalent reducton n lumnance of 1.5 to 2. log unts (Pastemak and Mergan, 1981). The acuty/lumnance curve s much shallower n the cat, and such a decrease n lumnance would result n only a.4 octave loss n acuty (Pastemak and Mergan, 1981). Indeed, the observed magntude of acuty loss n strobe-reared cats was close to that suggested by ths analyss.

6 The Journal of Neuroscence Vson Wthout Cortcal Drectonal Selectvty 943 Cat 81 Cat 84 Cat 812 /.28 c/d.28 c/d loo- 4 O lo- l * C/d I Drft Rate (Hz) Fgure 5. Contrast senstvty for detecton (open symbols) and drecton dscrmnaton (flled symbols) for the three strobe-reared cats. Numbers n the upper rght corner ndcate spatal frequency. Dscrmnaton data are not shown for.77 cycles/deg, snce at ths spatal frequency the cats could not dscrmnate drectons above 75% correct level. Orentaton dscrmnaton Orentaton dfference thresholds for low spatal frequency gratngs were unaffected by stroboscopc llumnaton. Ths result s consstent wth the normal number of orentaton-selectve neurons, as well as the normal wdth of orentaton tunng n area 17 of these cats (Cynader and Chemenko, 1976; Pastemak et al., 1985a). Although our measurements were lmted to a sngle spatal frequency of hgh contrast, t s unlkely that measurements at other spatal frequences or at lower suprathreshold contrasts would reveal abnormaltes n the orentaton response of the cats, snce orentaton thresholds are relatvely nsenstve to changes n spatal frequency and contrast (Bradley and Skottun, 1984; Burbeck and Regan, 1983; Caell et al., 1983). The present estmates of felne orentaton thresholds for gratngs are somewhat hgher than those reported for sngle lnes by some nvestgators (Hrsch, 1972; Vandenbussche and Orban, 1983; Wark and Peck, 1982; but see Berkley and Sprague, 1979). Several factors may account for the dfferences n thresholds between our estmates and those of others, ncludng stmulus confguraton (gratngs versus sngle lnes), the relatvely conservatve threshold crteron used here (75% correct), the relatvely low mean lumnance of the gratngs (3.8 cd/m*), and the fact that, n the present study, the two comparson stmul could not be vewed smultaneously. Temporal vson Measurements of crtcal flcker frequency and of the detecton of movng patterns revealed no sgnfcant abnormaltes n strobe-reared cats. Ther CFF ranged from 38 to 45 Hz, whch s nearly dentcal to that obtaned from normal cats and to the temporal resoluton data reported n other studes for ths lu- mnance level (e.g., Loop et al., 198). For both groups of cats, optmal temporal frequency (see detecton data n Fgs. 4 and 5) ranged from 2 to 5 Hz, dependng on spatal frequency. Ths result s consstent wth the optmal temporal frequences of Drft Rate (Hz) Fgure 6. Senstvty ratos (dscrmnaton/detecton) for normal (open symbols, cat 113) and strobe-reared (flled symbols) cats. Each pont for a normal cat s based on contrast senstvty data shown n Fgure 4; ratos for the three strobe-reared cats were calculated from the mean contrast senstvtes shown n Fgure 5. Spatal frequency s ndcated by numbers next to each curve. A rato of 1. ndcates that the eats requred the same contrast to detect the movng gratng and to dentfy ts drecton of moton.

7 944 Pasternak and Lenen Vol. 6, No. 4, Apr neurons n area 17 of both normal and strobe-reared cats (Movshon et al., 1978; Pasternak et al., 1985a). Moton thresholds and cortcal drectonal selectvty When the movng patterns were of hgh contrast, the deprved cats were able to dentfy the drecton of moton regardless of stmulus speed or confguraton. Even at the lowest detectable speed, the cats dscrmnated drectons as well as the normal cats. Ths result s surprsng, because less than 1% of ther cortcal neurons were drectonally selectve. There are several possble nterpretatons of ths fndng. Frst, the cats could have used cues other than stmulus drecton durng ths dscrmnaton. For example, sgnals from the eye movement system and/or phase dfferences between the two comparson movng dot or gratng patterns could also provde cues to drecton. A contrbuton of eye movements s dffcult to rule out, snce such patterns elct optoknetc nystagmus n both normal cats (Evnger and Fuchs, 1978) and n strobe-reared cats (Conway et al., 1981). However, t s unlkely that such a cue made any sgnfcant contrbuton to our fndngs. The contrast senstvty of strobe-reared cats for the detecton of.77 cycles/deg gratngs was wthn the normal range; thus, eye movements should also have been elcted by these patterns. Nevertheless, strobe-reared cats were unable to dscrmnate between the two drectons of moton at ths spatal frequency. Another bass for dscrmnatng drectons mght have been dfferences n local spatal features between the two comparson dot patterns. Ths, however, s also unlkely, snce dscrmnaton performance was largely unaffected by the use of perodc, hgh spatal frequency gratngs (see, also, Burr, 198). An alternatve explanaton of the present results s that vsual cortcal or subcortcal areas outsde of area 17 retaned normal numbers of drectonally selectve cells, and that these were nvolved n drecton dscrmnaton n strobe-reared cats. Spear et al. (1985) recently gathered data from the lateral suprasylvan (LS) area of some of the cats reared n 8 Hz lght and behavorally tested n our laboratory. They found that the LS, whch normally contans a hgh percentage of drectonally selectve cells (79%) was almost devod of such cells (8%) n these cats. Recordngs from the superor collculus of cats reared under smlar condtons have also shown a profound loss of drectonal selectvty (M. Cynader, personal communcaton). Furthermore, neurons n area 18 of cats reared under somewhat lower frequency stroboscopc llumnaton also show drastcally reduced drectonal selectvty (Kennedy and Orban, 1983). Thus, every cortcal area known to have sgnfcant numbers of drectonally selectve cells shows large losses of such neurons. Obvously, there remans the possblty that stll other vsual areas (not yet studed) contan normal numbers of cells that encode drecton and are capable of medatng drecton dscrmnatons. It s more lkely, however, that the few remanng drectonal cells n the vsual cortex are suffcent to sgnal dfferences n drecton of hgh-contrast movng patterns. Ths explanaton s consstent wth our fndng that, at low stmulus contrasts, drecton dscrmnaton n strobe-reared cats breaks down. Although the contrast senstvty of strobe-reared cats for detectng movng gratngs was only slghtly below normal, ther senstvty for dscrmnatng the drecton of gratng moton was severely depressed. It should be noted that these defcts were found at low spatal frequences, whch apparently were not affected physologcally by strobe-rearng (Pastemak et al., 1985a). Snce both the neural and behavoral defcts are largely lmted to condtons n whch the drecton of stmulus moton s vared, t s reasonable to assume that the profound and specfc defct n drecton dscrmnaton resulted from the loss of drectonally selectve neurons. Therefore, t s lkely that the resdual senstvty of strobe-reared cats for drecton was due to the lmted number and/or the reduced contrast senstvty of remanng neurons that encode drecton. The locaton of these neurons s not clear, snce the loss of drectonal cells s not lmted to area 17, but extends to the lateral suprasylvan area (Spear et al., 1985) and, most lkely, to area 18 (Kennedy and Orban, 1983). Physologcal studes of cat cortex show that neuronal mechansms processng low spatal and hgh temporal frequences are located manly n area 18, whereas those processng hgh spatal and low temporal frequences are found n area 17 (Berard et al., 1982; Bst et al., 1985; Movshon et al., 1978). Dfferences between neurons n these two cortcal areas have also emerged n an analyss of drectonal selectvty. Orban and colleagues (Orban et al., 1978, 1981 a, b) found a sgnfcantly larger proporton of drectonally selectve cells n area 18 and n area 17 n the regon devoted to central vson. Thus, neurons n area 17 appear to be more senstve to slowly movng, fne patterns, often rrespectve of stmulus drecton, and neurons n area 18 are more responsve to the drecton of moton of coarse patterns movng at hgher speeds. Snce the resdual senstvty for drecton n strobereared cats could be measured only n the spatal frequency range to whch neurons n area 17 of both normal and strobe-reared cats are relatvely nsenstve (Movshon et al., 1978; Pastemak et al., 1985a), the resdual senstvty of the strobe-reared cats to drecton may be determned by the remanng drectonally selectve neurons n area 18. At low spatal frequences, normal cats dentfed the drecton of the movng gratng at contrasts nearly dentcal to those requred for detecton. Dfferences between detecton and drecton dscrmnaton were pronounced at low speeds and hgher spatal frequences (Fgs. 4 and 6). Several nvestgators have compared the detecton and drecton dscrmnaton of human observers for movng snusodal gratngs (Lenne, 198; Mansfeld and Nachmas, 1981; Thompson, 1984; Watson et al., 198), and reported a smlar pattern of results: Dscrmnaton of the drecton of moton for slowly movng gratngs of hgh spatal frequency requres more contrast than s requred for smple detecton. Dfferent thresholds suggest that these two types of tasks depend on dfferent mechansms (Watson and Robson, 1981) and that the mechansms that detect hgh spatal and low temporal frequences do not encode drecton (Watson et al., 198). The large dfferences n senstvty between detecton and drecton dscrmnaton n cats wth nearly abolshed drectonal selectvty support ths nterpretaton. Snce the reduced senstvty for drecton of strobe-reared cats s most lkely due to the reduced number of drectonally selectve cells n ther vsual system, the low senstvty to drecton of normal cats at low temporal and hgh spatal frequences may also be ndcatve of a reduced nvolvement of drectonal neurons. Reduced senstvty to drecton wth such stmul may reflect a greater role of area 17 neurons and a decreased contrbuton of neurons n area 18, the majorty of whch are drectonally selectve and prefer coarse patterns movng at hgh veloctes. References Arend, L. E. (1976) Response of the human eye to spatally snusodal gratngs at varous exposure duratons. Vs. Res. 16: 131 l-l 315. Berard, N., S. Bst, A. Cattaneo, A. Forentn, and L. Maffe (1982) Correlaton between the preferred orentaton and spatal frequency of neurons n vsual areas 17 and 18 of the cat. J. Phvsol. _ (Land.) 323: Berkley, M. A., and J. M. Sprague (1979) Strate cortex and vsual acutv functons n the cat. J. Comu. Neurol. 187: Bst, S:, G. Carmngnoto, L. Gall, and L. Maffe (1985) Spatal frequency characterstcs of neurones of area 18 n the cat: Dependence on the velocty of the vsual stmulus. J. Physol. (Lond.) 359: Boothe, R. G., L. Korpes, and A. Hendrckson (1983) Ansometropc

8 The Journal of Neuroscence Vson Wthout Cortcal Drectonal Selectvty 945 amblyopa n Macaca nemestrna monkeys produced by atropnzaton of one eye durng development. Invest. Ophthalmol. Vs. Sc. 22: Bradley, A., and B. C. Skottun (1984) The effects of large orentaton and spatal frequency dfferences on spatal dscrmnatons. Vs. Res. 24: Burbeck, C. A., and D. Regan (1983) Independence of orentaton and sze n spatal dscrmnatons. J. Ophthalmol. Sot. Am. 73: 169 l Burr, D. C. (198) Senstvty to spatal phase. Vs. Res. 2: Caell, T., H. Brettel, I. Rentschler, and R. Hlz (1983) Dscrmnaton thresholds n the two-dmensonal spatal frequency doman. Vs. Res. 23: Conway, J. L., G. T. Tmberlake, and A. A. Skavensk (198 1) Oculomotor changes n cats reared wthout experencng contnuous retnal mage moton. Exp. Bran Res. 43: Cynader, M., and G. Chemenko (1976) Abolton of drectonal selectvty n the vsual cortex of the cat. Scence 193: Eggers, H. M., and C. Blakemore (1978) Physologcal bass of ansometropc amblyopa. Scence 21: Evnger, C., and A. F. Fuchs (1978) Saccadc, smooth pursut, and ontoknetc eve movements of the traned cat. J. Phvsol. (Land.) 285: 2b Graves, A. L., and K. L. Morse (1982) Ansometropa s assocated wth contrast senstvty loss n cats. Invest. Ophthalmol. Vs. Sc. (Suppl.) 22: 6. Hrsch, H. V. B. (1972) Vsual percepton n cats after envronmental surgery. Exp. Bran Res. 15: Hubel, D. H., and T. N. Wesel (1962) Receptve felds, bnocular nteracton and functonal archtecture n the cat s vsual cortex. J. Physol. (Lond.) 16: Kennedy, H., and G. A. Orban (1983) Response propertes of vsual cortcal neurons n cats reared n stroboscopc llumnatons. J. Neurophysol. 49: Lenne. P. ( 198) Perceptual sgns of parallel pathways. Phlos. Trans. R. Sot. L&td. (Bol.] 9: Loop, M. S., S. Petuchowsk, and D. C. Smth (198) Crtcal flcker fuson n normal and bnocularly deprved cats. Vs. Res. 2: Mansfeld, R. J. W., and J. Nachmas (198 1) Perceved drecton of moton under retnal stablzaton. Vs. Res. 21: Nachmas. J. (1967) Effect of exposure duraton on vsual contrast senstvty wth square-wave gratngs. J. Ophthalmol. Sot. Am. 57: Movshon, J. A., I. D. Thompson, and D. J. Tolhurst (1978) Spatal and temporal contrast senstvty of neurons n area 17 and 18 of the cat s vsual cortex. J. Physol. (Lond.) 283: Orban, G. A., H. Kennedy, and H. Maes (1978) Influence of eccentrcty on velocty characterstcs of area 18 neurones n the cat. Bran Res. >59: Pastemak, T., and W. H. Mergan (198) Movement detecton by cats: Invarance wth drecton and target confguraton. J. Comp. Physol. Psychol. 94: Pastemak, T., and W. H. Mergan (198 1) The lumnance dependence of snatal vson n the cat. Vs. Res. 21: 1333-l 339. PasteAak, T., and W. H. Mergan (1984) The effects of stmulus speed on drecton dscrmnatons. Vs. Res. 24: Pastemak, T., W. H. Mergan, D. G. Flood, and D. Zehl (1983) The role of area centrals n the spatal vson of the cat. Vs. Res. 23: Pastemak, T., R. A. Schumer, M. S. Gzz, and J. A. Movshon (1984) Abolton of cortcal drectonal selectvty mpars drecton dscrmnaton of cats. Sot. Neurosc. Abstr. 1: 799. Pastemak, T., R. A. Schumer, M. S. Gzz, and J. A. Movshon (1985a) Abolton of cortcal drectonal selectvty mpars vsual behavor n cats. Exp. Bran Res. 61: Pastemak, T., D. G. Flood, T. A. E&n, and W. H. Mergan (1985b) Selectve damage to large cells n the cat retno-genculate pathway. J. Neurosc. 5: Quck, R. F. (1974) A vector magntude model of contrast detecton. Kybemetk 16: Sekuler, R., A. Pantle, and E. Levnson (1978) Physologcal bass of moton percepton.. In Handbook of Physoogv, % ol. 8: Percepton, R. Held. H. W. Lebowtz. and H.-L. Teuber. eds.. I_- DD Snrna- - - er-verlag, Berln. Smth, E. L., R. S. Harwerth, and G. W. Magure (1983) Effects of chronc atropnzaton on vsual acuty n kttens. Behav. Bran Res. 7: Spear, P. D., and T. P. Baumann (1975) Receptve-feld characterstcs of sngle neurons n lateral suprasylvan vsual area of the cat. J. Neurophysol. 38: Spear, P. D., L. Tong, M. A. McCall, and T. Pastemak (1985) De- -velopmentally nduced loss of drecton selectve neurons n cat lateral sunrasvlvan vsual cortex. Dev. Bran Res. 2: Thompson, P. (1984) The codng of velocty of movement n the human vsual system. Vs. Res. 24: Tulunay-Keesey, U., and R. M. Jones (1976) The effect of mcromovements on the eye and exposure duraton on contrast senstvty. Vs. Res. 16: Vandenbussche, E., and G. A. Orban (1983) Merdonal varaton n the lne orentaton dscrmnaton of the cat. Behav. Bran Res. 9: Wark, R. C., and C. K. Peck (1982) Behavoral consequences of early vsual exposure to contours of a sngle orentaton. Dev. Bran Res. 5: Watson, A. B. (1979) Probablty summaton over tme. Vs. Res. 19: Watson, A. B., and J. G. Robson (198 1) Dscrmnaton at threshold: Labelled detectors n human vson. Vs. Res. 21: 1115-l 122. Orban, G. A., H. Kennedy, and H. Maes (198 la) Response to move- Watson, A. B., R. G. Thompson, B. J. Murphy, and J. Nachmas (198) ment of neurons n areas 17 and 18 of the cat: Drecton selectvty. Summaton and dscrmnaton of gratngs movng n opposte d- J. Neurophysol. 45: rectons. Vs. Res. 2: Orban, G. A., H. Kennedy, and H. Maes (198 1 b) Response to move- Wlson, H. R. (1985) A model for drecton selectvty n threshold ment of neurons n areas 17 and 18 of the cat: Velocty senstvty. moton percepton. Bol. Cybem. 51: J. Neurophysol. 45: Pastemak, T., and L. Lenen (1984) Drecton dscrmnaton n strobereared cats. Invest. Ophthalmol. Vs. Sc. (Suppl.) 25: 217.