Receptive Fields of Visual Neurons in Primary Auditory Cortex

The Journal of Neuroscience, September 1992, 12(g): 36513664

Visual Projections Routed to the Auditory Pathway in Ferrets: Receptive Fields of Visual Neurons in Primary Auditory Cortex

Anna W. Roe,a Sarah L. Pallaqb Young H. Kwon, and Mriganka Sur Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

How does cortex that normally processes inputs from one ulate nucleus, or MGN). Retinal afferents subsequently provide sensory modality respond when provided with input from a visual input to cells in primary auditory cortex (Al) of the different modality? We have addressed such a question with “rewired” ferret (Sur et al., 1988) and establish a topographic an experimental preparation in which retinal input is routed visual map there (Roe et al., 1990a). We have now examined to the auditory pathway in ferrets. Following neonatal sur- both qualitatively and quantitatively the physiological response gical manipulations, a specific population of retinal ganglion properties of single visual units in Al of rewired ferrets and cells is induced to innervate the auditory thalamus and pro- compared them to the properties ofcells in primary visual cortex vides visual input to cells in auditory cortex (Sur et al., 1988). (V 1) of normal ferrets. We have now examined in detail the visual response prop- A preliminary report of these data has been published pre- erties of single cells in primary auditory cortex (Al) of these viously (Roe et al., 1990b). rewired animals and compared the responses to those in primary visual cortex (Vl) of normal animals. Cells in Al of Materials and Methods rewired animals differed from cells in normal Vl: they ex- Neonatal surgery to route retinal projections to the MGN hibited larger receptive field sizes and poorer visual re- To induce retinal innervation of nonvisual thalamic structures, it is sponsivity, and responded with longer latencies to electrical necessary to remove the normal retinal targets [primarily the lateral geniculate nucleus (LGN) and the superior colliculus (SC)] and to make stimulation of their inputs. However, striking similarities were alternative target space available [in this case, the medial geniculate also found. Like cells in normal Vl, Al cells in rewired an- nucleus (MGN); see Schneider, 19731. imals exhibited orientation and direction selectivity and had Procedures were similar to those described previously (Pallas et al., simple and complex receptive field organizations. Further- 1990). Briefly, timed pregnant pigmented ferret jills were bred in our colony or purchased from a commercial supplier (Marshall Farms). more, the degree of orientation and directional selectivity as Within 24 hr of parturition, ferret kits were removed and anesthetized well as the proportions of simple, complex, and nonoriented by deep hypothermia. All surgery was performed under sterile condi- cells found in Al and Vl were very similar. These results tions. A 1 cm midline incision was made in the scalp and the skull was have significant implications for possible commonalities in exposed. After removing a small piece of overlying skull, the SC was ablated by cautery. To induce massive retrograde degeneration of the intracortical processing circuits between sensory cortices, LGN, large regions of visual cortex (including areas 17, 18, and often and for the role of inputs in specifying intracortical circuitry. 19) were lesioned by cauterizing through the skull. Fibers running in the brachium of the inferior colliculus were transected at the mid-SC Is cortex of a given sensory modality uniquely specified to pro- level, thereby denervating the MGN. All lesions were made unilaterally cess inputs of that modality, or can it, under certain conditions, (bilateral lesions significantly reduced survival rates). On completion of process input of another modality? The question is important surgery, the skin was sutured and the kit revived under a heat lamp. A single dose of antibiotic (0.01 cc amoxycillin, 100 mg/ml) was given for understanding intrinsic and extrinsic determinants of cor- subcutaneously, and antibacterial ointment was applied to the wound. tical development, and for understanding plasticity and recovery The kit was then returned to the mother for rearing. of function following early brain trauma. In this article, we have These lesions result in induction of developing retinal afferents into addressed the issue by routing visual inputs into the auditory the MGN (Sur et al., 1988). Since the normal thalamocortical connec- pathway during development and subsequently studying the tivity from the MGN to primary auditory cortex (Al) is retained (Pallas et al., 1990), an aberrant visual pathway from the retina to MGN to resulting visual responses of cells in auditory cortex in the adult. Al is established (Sur et al., 1988). We have previously demonstrated that, following specific neonatal lesions, retinal afferents in the ferret can be induced to Visual physiology in adult ferrets innervate the auditory thalamus (specifically, the medial genic- Surgicalpreparation. Normal and rewired adult ferrets were prepared for visual physiology using procedures similar to those described pre- Received Jan. 6, 1992; revised Apr. 14, 1992; accepted Apr. 20, 1992. viously (Roe et al., 1989). Animals were anesthetized, paralyzed, and We thank Preston Garraghty for participating in early experiments and for respirated. Following induction of anesthesia with ketamine hydro- critical review of an earlier version of the manuscript, Sacha Nelson for reviewing chloride (30 mg/kg) and xylazine (2 mg’kg), a cannula (24 gauge) for the manuscript, and Terry Sullivan for excellent technical assistance. This work anesthetic and paralytic delivery was implanted in either the femoral was supported by a Whitaker Foundation fellowship (A.W.R.), NIH Postdoctoral or the jugular vein, and an endotracheotomy was performed. Animals Fellowship EY 06 12 1 (S.L.P.), NIH Grant EY 077 19 (MS.), and a grant from the McKnieht Foundation (MS.). were then placed in a stereotaxic apparatus, paralyzed, and artificially = Pres&t address, lo whichcorrespondence should be addressed: Anna W. Roe, respired (rate, 30dO/min; vol, 25-30 cc) with a 70:30 mixture ofnitrous Laboratory of Neurobiology, Rockefeller University, 1230 York Avenue, New oxide and oxygen. Ferrets received a constant intravenous infusion of York, NY- 1002 I. a 5% dextrose solution containing ketamine hydrochloride (10 mg/kg/ h Present address: Division of Neuroscience, Baylor College of Medicine, I hr) for anesthesia and gallamine hydrochloride (3.6 mg/hr/kg) for mus- Baylor Plaza, Houston, TX 77030. cular paralysis. End-tidal CO, was maintained at 4.0%, heart rate was Copyright 0 1992 Society for Neuroscience 0270-6474/92/12365 l-14$05.00/0 monitored, and body temperature was maintained at 38°C with a heating 3652 Roe et al. * Visual Responses from Auditory Cortex of Rewired Ferrets

Table 1. Number of units recorded in VI and Al

Vl Al Total 153 495 Electrically driven 124 (95) 485 (98) J Visually driven I05 (69) I94 (39) E 60 Visually and electrically driven 77 (59P I84 (37) 8 Receptive field size determined 101 (66) 100 (20) t n Ocular dominance determined 76 (50) 110 (22) 40 Cell class determined IO1 (66) 52 (11) Quantitatively characterized 38 (25) 24 (5) Driven by auditory stimuli 0 (0) 0 (0) 20 Percentageof total is shown in parentheses. ” Percentage calculated from a total of I3 I cells for which chiasm latencies were assessed. In one experiment, chiasm latencies were not assessed (22 cells). 0 u da 1 2 3 4 pad. Supplemental halothane (1%) was supplied during all surgical pro- Responsivity cedures. The skull and dura overlying visual cortex in normal animals and auditory cortex in rewired animals were removed. Stimulating elec- Figure 1. Distribution of responsivities (1 indicates poor responsive- trodes were lowered into the optic chiasm and cemented in place. Drops ness and 4 good responsiveness) of cells in A I (dark bars) and V 1 (light of atropine sulfate and phenylephrine hydrochloride were applied to the bars). These distributions are significantly different 0, < 0.000 I, Mann- eyes to dilate the pupils. Eyes were fitted with zero-power contact lens Whitney I/ test). Out of a total of 194 cells, 38 recorded from early and focused on a tangent screen I I4 cm in front of the animal. Optic disks were plotted by reflection onto the tangent screen. experiments in rewired animals were not rated for responsivity. Recording and data collection. In I8 rewired and 13 normal adult ferrets, we used parylene-insulated tungsten microelectrodes (2-3 M0 impedances) to record extracellularly from isolated cells in A 1 of rewired and Vl of normal animals. Recording penetrations were made in Al at 50 Frn and stained with cresyl violet. Lesions and recording tracks (located between the anterior and posterior limbs of the ectosylvian were identified and located with respect to the borders of A I sulcus; Kelly et al., 1986) and V 1 (located in the caudal half of the lateral Data analysis. For each cell that was studied quantitatively, peak sulcus: Law et al., 1988) on the basis of sulcal patterns. Penetrations response (maximum spikes/set summed over at least 10 trials) minus were usually spaced 250-500 pm apart. Unit activity was assessed every background (average spikes/set of spontaneous activity summed over 50-100 Km in each penetration (1) for response to electrical stimulation identical number of trials) was calculated for every stimulus condition. of the optic chiasm (l-l 5 V) and (2) for response to visual stimulation. From these, orientation, velocity, and bar size tuning curves were plotted In two rewired animals and one normal animal, pure tone bursts (ranging and selectivity indices calculated. The orientation index was defined as from 8 to 20 kHz) or broad-band auditory stimuli (100 ms square-wave I - (orthogonal response/best response), and the direction selectivity clicks or white noise) were delivered binaurally through earphones. Vi- index was defined as I - (response to direction opposite of preferred sually responsive units were plotted and characterized on the tangent direction/response to preferred direction). These are normalized indices, screen with a hand-held projection lamp. In addition to receptive field with 1.O indicating a highly selective and 0 a nonselective cell. In ad- location (as determined by reflection of the optic disk, located at 33” dition, orientation widths were determined at l/d height (cf. Schiller azimuth and ~ 3” elevation; Zahs and Stryker, 1985) and receptive field et al., 1976b). Unless otherwise specified, statistical tests between two size (measured as I” per 2 cm on a tangent screen located 1 I4 cm from parameters employed at the Mann-Whitney I/ test. Comparison of the eye), a number ofother response characteristics were also examined. groups was done using the x’ test. These included ocular dominance; orientation, direction, and velocity selectivities; spatial organization of on- and off-responses within the receptive field; presence or absence of summation and end-stopping; Results and strength of response. When possible, oriented cells were classified as simple or complex (Hubel and Wiesel, 1962). Electrolytic lesions (5 Four-hundred and ninety-five visual units were recorded from PA for 5 set) were made in selected penetrations. A 1 of rewired ferrets and 153 visual units from V 1 of normal The tuning characteristics of a subset of isolated single units were ferrets (see Table 1). Nearly all cells responded to electrical studied quantitatively. For quantitative determination of orientation, stimulation of the optic chiasm. Of these, 194 (39%) of the cells direction, and velocity tuning preferences, moving bar stimuli, either projected with an optic bench or generated on a Tektronix monitor in Al and 105 (69%) of the cells in VI were visually responsive driven by a PICASSO (Innisfree) stimulus generator, were continuously under our recording conditions. The smaller overall proportion swept (without interleaving) at each orientation (usually at 30” intervals) (pooled over all animals studied) of visual cells recorded in A I and velocity (from S”/sec to 100”/sec) over the receptive field; each may be due to poorer responsiveness of A 1 units (see below) as successive orientation and velocity was presented in a pseudorandom well as to interanimal variability in lesion size. [In some rewired fashion. Spike responses of the cell under study were isolated with a pulse-shape discriminator (Ealing). For each category of stimulus, 15 animals with large visual cortical lesions, almost all units re- cycles of response were collected and summed by the UNKELSCOPE pro- sponsive to chiasm stimulation were visually driven, while in gram run on an IBM PC286 computer, and poststimulus time (PST) animals with small cortical lesions a much lower percentage of histograms were generated on line. Spike responses were also stored on visual units was encountered. This variability probably relates tape, and PST histograms were analyzed off line. Background activity was collected in the absence of visual stimulation. to the size of the extant LGN, and hence to the size of the Histology. Following the completion of data collection, the animal rerouted retinal projection to the MGN (Roe, 1991).] A subset was killed with an intravenous or intraperitoneal overdose of pento- of the visual cells in rewired A 1 and normal VI were charac- barbital (65 mg/kg) and perfused through the heart with saline followed terized in further detail and their responses studied quantita- by fixative (1% paraformaldehyde, 2% glutaraldehyde) and 10% and 20% sucrose solutions. The brain was then removed from the cranium tively (Table 1). Of the units that were tested for auditory re- and stored in 30% sucrose overnight. After brain sulcal patterns (Kelly sponsiveness (n = 35 in Al of two rewired ferrets; n = 22 in et al., 1986) were recorded photographically, frozen sections were cut V 1 of one normal ferret), none responded either to clicks or to The Journal of Neuroscience, September 1992, i2(9) 3853

white noise. These sameauditory stimuli produced robust re- A 4o sponsesin normal Al (n = 37 in one normal ferret). We have not attempted to conduct a laminar analysisof our data. However, visual units were recorded throughout cortical layers in both Al and Vl: cells were recorded throughout the = 1 mm thickness of ferret A 1 (Roe, 1991).

Differences between response features of cells in Al and Vl There were three major differencesbetween the characteristics of visual units in Al and Vl: responsivity, receptive field size, and latency to optic chiasm stimulation. As will be discussed later, thesedifferences appear to arise primarily from the properties of retinal and thalamic input to A 1 and V 1.

Responsivity One salient difference between visual units in Al and Vl was their responsivenessto visual stimulation. Most cells in VI respondedin robust and consistent fashion to visual stimuli. In contrast, cells in A 1 were much more difficult to drive, exhibited labile responses,and responded with fewer spikes. Cells were B 40- qualitatively rated, during recording, on a four-point scale,where 1 indicates poor responsivenessand 4 good responsiveness.This . . 20 - l * . . . index reflects both robustness(qualitative determination of re- c sponsestrength, i.e., number of spikes)as well as reliability of response.(Units that were responsive only to electrical stimu- .-s o- lation were not rated; 38 Al cells recorded in early experiments 5 . 0 also were not rated.) Of 105 V 1 cells and 156 A 1 cells that were 2 . rated, 93% of units in Vl were rated 4 and 68% of units in Al w -2o- . were rated 1 (Fig. 1). The poorer responsivity of Al cells often made receptive field characterization of Al cells difficult and -4o- time consuming; proportionally fewer cells in Al were fully . characterized as a result (seeTable 1). .fl*.. . . -60 I I I I 1 Receptive field size -20 0 20 40 60 60 A second prominent difference between A 1 and V 1 cells was Azimuth (“) the size of their receptive fields. Receptive field sizes(calculated as the averageof the length and width of the receptive field) of visual cells in A 1 were significantly larger than thoseof V 1 cells, their mean diameter (mean = 10.9”, n = 87) reaching almost three times that of Vl cells (mean = 3.9”, n = 101) (Fig. 2A). Of the cells in A 1,64% had receptive fields larger than the largest receptive fields recorded in V 1. In addition, 13% of cells in A 1 had receptive fields that were large (15-30” in diameter) and whose borders were difficult to determine precisely (Fig. 2‘4, large, diffuse; not included in calculation of mean). The difference in the receptive field sizeswere not eccentricity related. The receptive field eccentricities of cells recordedin Al were significantly different from those of cells recorded in Vl (p < 0.0001, Mann-Whitney U test) (Fig. 2B), consistent with the much flatter distribution of visual field magnification in Al compared to Vl (Roe et al., 1990a). However, receptive field sizes in Al were still larger (p < 0.0001) when eccentricity- matched (~30”) populations were compared (Fig. 20. This was Receptive Field Diameter (“) also true for eccentricity-matched samplesbelow 15” (A 1, n =

Figure 2. A, Receptivefield diameters(defined as averageof width and heightof receptivefield) of cells in Al (dark bars) and Vl (light bars). Receptivefield sizesin Al (mean= 10.9”, solid arrowhead) are significantlylarger (p < 0.0001)than thosein Vl (mean= 3.9”, open than in Vl @ < 0.0001).C, Takinginto considerationonly receptive arrowheud). B, Distributionof receptivefield locationsrecorded in Al fields whoseeccentricities are lessthan 30” revealsthat Al cellsstill (so/id circles) and VI (open squares). In Al, significantlymany more have much largerreceptive fields (Al mean= 9.7”, Vl mean= 3.9”,p fieldswere recorded representing peripheral regions of the visual field < 0.0001). 3654 Roe et al. * Visual Responses from Auditory Cortex of Rewired Ferrets

Latency (msec)

Figure 3. Latencies of cells in Al and V 1 to electrical stimulation of the optic chiasm. A, Cells in Al (dark bars, n = 485, mean = 6.8 msec) have significantly longer latencies to chiasm stimulation than those in VI (light bars, n = 124, mean = 4.4 msec) @ < 0.0001). B, Jatencies of visually responsive cells in A 1 (s/x&~ bars, n = 184, mean = 6.9 msec) are similar to those ofall cells recorded in A I (p > 0.13). C, Similarly, visually responsive cells in VI (shaded bars, n = 77, mean = 4.3 msec) have latencies similar to all cells recorded in VI @ > 0.7). Latency (msec) The Journal of Neuroscience, September 1992, 12(9) 3655

18; Vl, n = 53; p < 0.0001) and below 45” (Al, n = 78; Vl, n 80 = 101; p < 0.0001). 1

Conduction latency qf qferents 60 - q Vl (n=76) Cells in A 1 and V 1 also differed in their latencies to optic chiasm E In Al (n=llO) stimulation. When the latencies of all cells in Al and Vl were i3 40- compared (Fig. 3A), those recorded in Al (mean = 6.8 msec) f were significantly longer than those in Vl (mean = 4.4 msec; p n < 0.0001). In both Al (Fig. 3B) and Vl (Fig. 30, the latencies 20 - recorded for the subpopulation of visually responsive cells (A 1, n = 194; V 1, n = 77) were representative of all units recorded. There was no correlation between the latency to optic chiasm 0 stimulation and cortical depth, receptive field size, or responsivity. 1234567 Ocular Dominance Similarities between cells in Al and Vl: integrative features of receptive fields Figure 4. Ocular dominance distributions of visual cells recorded in Al (dark bars) and in Vl (light bars). These two distributions are not Of the visual cells recorded in Al of rewired ferrets, a portion significantly different [x*(0.95) = 4.6, df = 61. were sufficiently responsive to allow further characterization, including determination of cell class, ocular dominance, and orientation, direction, and velocity selectivities. As mentioned Receptive,fild types previously, reliably responding visual cells were much more VI cells. Receptive fields of cells in Vl of normal ferrets were frequently encountered in Vl than in Al; this difference

60 1 An example of an orientation selective cell in V 1 is shown in Fig. 6A. This cell is a simple cell and responds optimally to a bar oriented at -45” from vertical, sweeping in either direction. A second example is shown in Fig. 6B. This complex cell prefers Cl Vl (n=lOl) bars oriented at +45” from vertical and is slightly directional W Al (n=52) for upward movement. Cells in Al were also characterized similarly. Fig. 7A illus- trates the orientation selectivity of a nondirectional simple cell in A 1 that preferred horizontally oriented bars. A directionally selective simple cell is shown in Fig. 7B. This cell responds best to downward-sweeping bars oriented 30” from vertical. As can be seen from these plots, cells in Al respond with fewer spikes overall than those in Vl and their responses appear to be modulated less strongly with respect to background levels. Orientation tuning. The orientation tuning of 38 cells in Vl and 24 cells in Al was studied quantitatively. For each of these cells, two measures of orientation tuning were calculated: orientation selectivity and orientation tuning width. The orienta- simple complex non- k tion selectivity index is defined as: 1 - (orthogonal response/ oriented best response) (cf. Felleman and Van Essen, 1987). A cell whose Class response in the best orientation is twice the level of that to the Figure 5. Percentageof simple, complex, and nonoriented cells in Al orthogonal orientation would have an orientation selectivity of (dark bars) and VI (light bars). These distributions are not significantly 0.5; one whose best response is 1.5 times that in the orthogonal different [x2(0.95) = 3.7, df = 21. orientation would have a selectivity of0.33. We have considered those units with selectivities below 0.3 to be unoriented, those with selectivities from 0.3 to 0.5 to be weakly oriented, and and responded equally well to small and large bars. End-stop- those with selectivities above 0.5 to be strongly oriented. Both ping was seen in one simple and five complex cells. In one case Al and Vl contained cells exhibiting a range of orientation (a simple cell), the degree of end-inhibition was quite dramatic: tuning selectivities, ranging from unoriented to strongly orient- extending a light bar by 10% beyond the receptive field border ed. In fact, no significant difference was seen in the distribution reduced the cell’s response by 50%. Nonoriented receptive fields of orientation strengths of cells in Al and Vl (Fig. 8A). Of (n = 12) were either circular or elongated with weak or absent oriented cells, approximately 40% of units in Al (three simple, inhibitory surrounds. six complex) and VI (six simple, eight complex) were sharply Relative proportions. Not only do simple, complex, and nonoriented (selectivity > 0.5) while 33% of units in Al (five sim- oriented cell types occur in A 1, they also occur in proportions ple, three complex) and 26% in V 1 (three simple, seven complex) similar to that found in Vl. In both the Al and the Vl popu- were broadly oriented (selectivity 0.3-0.5). lations recorded, approximately 50% of the cells were complex, Orientation tuning widths (defined as the width of the ori- 25% simple, and 25% nonoriented (Fig. 5). While we do not entation tuning curve at l/\/2 peak height) were also deter- claim that these figures reflect actual cell class composition in mined. For both Al and VI cells, orientation tuning widths Al and V 1, at no time did we attempt to “find” cells of any ranged from narrow tuning (< 30”) to very broad tuning (> 90”) particular class. Our sampling was based solely on units with (Fig. 8B). While a greater percentage of cells in Vl than in Al appropriate strength and consistency of response. We found no had tuning widths of 30”-59”, these distributions were not sig- statistically significant correlations, in either A 1 or V 1, between nificantly different. Thus, neither measure of orientation tuning receptive field type and receptive field size, latency to optic revealed any quantitative difference between Al and VI with chiasm stimulation, or depth. Neither were there any correla- respect to orientation tuning. tions between cell type and orientation, direction, or velocity Direction selectivity. A directionality index was calculated from selectivities (see also below). the responses obtained at the preferred orientation of each cell; this index was defined as 1 - (nonpreferredipreferred response). Tuning characteristics of single units in Al and VI The frequency distribution of these values in Al and Vl also To determine orientation tuning, we first examined responses proved to be similar (Fig. 9). Interestingly, V 1 contained a higher to flashing bars at various orientations. For cells not responsive proportion (40%) of nondirectional (directional selectivity < to standing flashes, we made a qualitative assessment of ori- 0.3) cells than Al (2 1%). A 1 also contained a greater proportion entation selectivity with moving or “jiggling” bars or moving (25%) of highly directional cells (selectivity > 0.6) than VI edges. We also tested the response to moving spots; however, (15%). Overall, however, the two distributions were not signif- moving spots were generally not an effective stimulus. For quan- icantly different (0.1 < p < 0.5). No relationship was seen be- titative determination of orientation selectivities, moving bars tween cell class and directionality in either Al or Vl: direc- were used as stimuli and polar plots made of the responses. In tionality strengths for each of the simple, complex, and almost all cases, quantitative characterization of the optimal nonoriented cells classes ranged from nondirectional to highly orientation agreed with our qualitative assessment. For some directional. cells in A 1, quantitative collection of spikes in response to mul- Velocity selectivity. One of the more unexpected findings of tiple stimulus sweeps revealed orientation tuning not obvious this study was that the velocity tuning ofcells in Al were similar from responses to single sweeps assessed by ear. to those of cells in Vl. The velocity selectivity of each cell was Figure 6. Orientation tuning of two cells in Vl: polar plots of each cell’s response (peak minus background) to sweeping bars at different orientations and directions (indicated by bars with arrows). PST histograms for each bar orientation and direction are shown. A, A simple ceI1 with an orientation tuning selectivity of 0.57, an orientation tuning width of 63”, and a directionality selectivity of 0.25. Stimulus bar, 2.4” x 4.4” (L x W), sweeping at 1Y/set. B, An end-stopped complex cell with an orientation tuning selectivity of 0.73, an orientation tuning width of 65”, and a directionality selectivity of 0.38. Stimulus bar, 6.1” x 3.0” (L x W), sweeping at 12”/sec. Figure 7. Orientation tuning of two cells in Al (conventions as in Fig. 6). A, An end-stopped complex cell with orientation tuning selectivity 0.73, orientation tuning width 659 and directionality selectivity 0.38. Stimulus bar, 5.7 x x 6.5” (L x W), sweeping at 72”/sec. B, A simple cell with orientation tuning selectivity 0.63, orientation tuning width 47”, and directionality selectivity 0.7 1. Stimulus bar, 12.5” x 13.2” (L x W), sweeping at 78”lsec. The Journal of Neuroscience, September 1992, 72(9) 3659

~~A 3o VI~~

~~I I~~

~~0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.6 0.9 1.0 0 Orientation Selectivity 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Directionality~~

Figure 9. Directionality of cells in Al (dark bars) and in VI (fight B 5o ; bars). Al, mean = 0.45; VI, mean = 0.36. The two distributions are 1 n not significantly different (0.1 < p < 0.5). 401 II 10 Vl (n=38) 1 Discussion Our goal in this study has been to explore the developmental and functional specificity of mammalian cerebral cortex. To- ward this goal, we have challenged Al in ferrets with visual input and studied the resulting visual processing in A 1. Previ- ously, we reported that retinal projections can be induced to innervate auditory thalamus in ferrets following specific neonatal lesions (Sur et al., 1988; cf. Schneider, 1973). These projections, which appear to arise primarily from retinal ganglion cells of the W cell class, provide visual input to cells in A 1 (Sur Orientation Width (“) et al., 1988; Pallas et al., 1990; Roe et al., 1990a; Roe, 199 1). In this article we have characterized in detail the receptive field Figure 8. Orientation tuning of cells in Al (dark bars) and in V 1 (light properties of visually responsive cells in Al of rewired ferrets bars). A, Distributions of orientation tuning selectivities in Al (mean and have compared them both qualitatively and quantitatively = 0.41) and Vl (mean = 0.39) are similar @ > 0.65). B, Distributions to visual cortical cells recorded in Vl of normal ferrets. We find of orientation tuning widths (defined as width at l/fl height) in Al (mean = 65.3”) and VI (mean = 72.5”) are not different @ > 0.90). that receptive field properties of visual cells in rewired A 1 bear remarkable similarities to those in normal Vl. We will first summarize and discuss the generation of these receptive field properties in Al and then turn to two possible interpretations studied by collecting responses to optimally oriented light stim- of these results. uli swept at several velocities, generally between Y/set and 1OO”/ set, over its receptive field; 29 Vl cells (9 simple, 11 complex, Generation of visual receptive field properties in Al and 9 nonoriented) and 18 A 1 cells (6 simple, 9 complex, and Sources of visual input to AI 3 nonoriented) were studied in this fashion. The stimulus ori- It is important to consider at the outset what the sources of entation chosen for nonoriented cells was arbitrary. visual input to Al are in rewired ferrets. It is possible,for ex- Examples of velocity tuning preferences are shown in Fig. ample, that there are novel corticocortical projections to Al in 1OA (A 1, right column; V 1, left column) and are similar to those rewired animals that arise from remaining unlesioned visual previously described in cat visual cortex (e.g., Orban et al., cortical areas, and that certain features of visual integration 198 1). Most cells were tuned to either slow (< 1S’/sec, curves 1 (such as orientation selectivity, direction selectivity, and bin- and 2), medium (15-50”/sec, curves 3), or fast (> 50”/sec, curves ocularity) are already present in thesecortical inputs. The issue 4) velocities. Some cells exhibited velocity high-pass or low- has been addressedby making injections of retrograde tracers pass characteristics (curves 5; see also curves 1). We recorded into Al in rewired ferrets, and comparing the cortical and tha- a few cells that responded equally well to slow and to fast, but lamic inputs with those to Al in normal ferrets (Pallas et al., not to medium, stimulus velocities (curves 6). Cells exhibiting 1988, 1990). In brief, the cortical areas that project to Al in broad tuning were also encountered (curves 7). Thus, cells in rewired animals are the sameareas that project to Al in normal both Al and Vl exhibited a range of velocity tuning. Further- animals, and do not include any visual cortical areas. more, the velocity tuning distributions were found to be similar The overwhelming proportion of thalamic input to Al in in these two areas: approximately 45% of the cells recorded in rewired animals arises from the MGN (Pallas et al., 1990). In Al and Vl were best tuned to slow, 35% to medium, and less addition, a sparseinput arises from the lateral posterior (LP)/ than 10% to fast velocities (Fig. 1 OB). No consistent relationship pulvinar complex to Al in rewired ferrets (Pallas et al., 1990). was seen between velocity tuning and cell type. It is unlikely that these LP/pulvinar projections contribute to 3660 Roe et al. - Visual Responses from Auditory Cortex of Rewired Ferrets

Figure 10. A, Examples of velocity tuning curves in V 1 (left column) and A Al (right column). In each column, curves I and 2 are tuned for slow (< 15”/ set), curves 3 for medium (15-50%ec), 1 and curves 4 for fast (>50%ec) velocities. Curves 5 display high-pass or low- FW6002 F695903 pass, curves 6 U-shaped, and curves 7 broad-band velocity characteristics. B (oppositepage), Distributions ofcells in A 1 (dark bars) and V 1 (light bars) displaying peak velocity preferences near 404. "'.""* 6%ec, 13”/sec, 25%ec, 50%ec, 1Oo”/ *oo" 20 40 60 60 100 set, as well as those displaying broad- band (BB) and U-shaped (slow/fast) 2 1 tuning curves. These distributions are 150 not significantly different (p > 0.8). Al, F906302 F906402 mean = 27.9%ec; V 1, mean = 24.6”/ sec. 100

~~50 2 0 20 40 60 60 100 120 3~~

~~F9016Ol~~

~~o- 0 20 40 60 60 100 4~~

~~loo- lo- 1 10 100 1000 0 20 40 60 60 100 24- 5 22' F695939 20.~~

~~..-\ SO- 161 0 20 40 60 80 100 120 160 6 2o 4o 6o 8o loo~~

~~0 FWO307 $ 100 r 8~~

~~0 b 20 40 60 60 20( 7 f F901602 F&s906 3iQ) lo< 20 -Y ::~~

~~:‘\ h 10 1 10 100 1000 0 20 40 60 80 100 Velocity (7s) Velocity (o/s) The Journal of Neuroscience, September 1992, f2(9) 3661~~

cells in the MGN. Visual cells in the MGN have circular or B 401 elongated receptive fields; thesecells lack orientation selectivity and exhibit lower degreesof direction selectivity than visual cells in A 1 (Sur et al., 1988; Roe, 1991). These findings parallel patterns of receptive field generation in the normal visual pathway through the LGN to area 17 in the ferret (Esguerraet al., 1986; Price and Morgan, 1987; Roe et al., 1989). While a limited degreeof orientation bias may be present at the retinal or thalamic level (e.g., Leventhal et al., 1990), the generation of a marked degree of orientation and direction selectivity is be- lieved to occur intracortically (e.g., Sillito, 1975). Direction selectivity is affected following selective lesionsof superficial cor-

0 tical layers (Eysel et al., 1987). These cortical receptive field 6 13 25 50 100 BB slow/fast properties of VI and Al cells are thus likely to arise from intracortical circuits common to both V 1 and visually innervated Best Velocity (o/s) A 1. In our recordings from normal V 1 as well as from rewired Figure IO. Continued. A 1, we found a fairly large proportion (= 25%) of nonoriented cells (Fig. 5). There appearsto be a true speciesdifference between ferret and cat visual cortex in terms of the prevalence of the visual responsesof Al neurons since (1) theseprojections nonoriented cells.While visual cortex in cats (Hubel and Wiesel, are much sparserthan the projections from the MGN, (2) they 1962) and mink (LeVay et al., 1987) has few nonoriented cells, arise from cells that lie deep within the LP/pulvinar region and approximately one-quarter of cells recorded in ferret Vl are that show almost no overlap with retinal projections in these nonoriented. A previous study of receptive fields in ferret Vl animals, and (3) ibotenic acid lesionsof the MGN abolish pre- reported only that “most single neurons . . . were orientation- viously visually responsive recording sites in Al and similar selective” (Law et al., 1988). It is unlikely that our nonoriented lesionsof the LP/pulvinar complex do not (S. L. Pallas,L. Toth, cells representrecordings from geniculocortical afferents, since and M. Sur, unpublished observations). (1) afferents were recognized by their higher level of spike activity, (2) nonoriented units were clearly distinguished as cell Contribution of retinal inputs spikes rather than axon spikes, and (3) nonoriented units were Our results show that visual cells in Al are poorly responsive recorded at all depths in the cortex. While it is possible that and have large receptive field sizes and long latencies to optic someofthe recording siteswere influenced by multiunit activity, chiasm stimulation. These characteristics are already present in the presenceof orientation selectivity would still be detected at and can be predicted by the retinal and thalamic input to Al. such sites since visual cortical cells within a single columnar We have suggestedpreviously that the auditory thalamus in locale sharesimilar orientation selectivities (LeVay et al., 1987; rewired ferrets receivesretinal ganglion cell input of the W cell Law et al., 1988; Redies et al., 1990). type (Roe et al., 1987; Sur et al., 1988; Pallas et al., 1989; Roe, The marked degree of similarity between Al and Vl is ap- 1991). Retinal W cells have “sluggish” responsivenessto visual parent not only in the orientation, direction, and velocity se- stimuli and are characterized by large, often diffuse, receptive lectivity of cells but also in the types of receptive field organi- fields, some of which have center-surround organization but zations found there. As in V 1, A 1 contains simpleand complex many of which do not (Wilson et al., 1976; Fukuda et al., 1984; (and nonoriented) receptive field types and in roughly similar Stanford, 1987). Similar receptive field properties are found in proportions. These findings are somewhatsurprising in view of postsynaptic visual cells in the MGN in rewired ferrets (Sur et the fact that Al and Vl receive significantly different types of al., 1988; Roe, 199 1). Large receptive field sizesof Al cells may input from the retina. V 1 in normal cats and ferrets is dominated be due to both the W cells that form the retinal sourceof inputs by X and Y cell inputs from the retina relayed through the LGN. and to the pattern of highly convergent and divergent thala- Previous studies devoted to examining the contribution of X mocortical connectivity typical of the auditory systemand shown and Y cell inputs to receptive field properties of simple and to be retained in these rewired ferrets (Pallas et al., 1990; cf. complex cells in primary visual cortex have shown that neither Andersen et al., 1980; Middlebrooks and Zook, 1983). The poor the simple nor the complex cell classesare exclusively X- or responsivity of cells in Al is consistent with its W type input Y-cell driven (Malpeli et al., 1981; Tanaka, 1983; Mullikin et (Sur and Sherman, 1982). In addition, we cannot rule out ab- al., 1984; Ferster 1990a,b). Given that visual inputs to Al normal synaptic configurations of retinal axons on MGN cells arise from retinal W cells (Roe, 1991), the results of this study or additional inhibitory mechanismswithin MGN or Al that demonstrate that (1) X and Y cell inputs are not necessary,and decreasethe responsivenessof MGN and Al cells in rewired (2) W cell input is sufficient, for the generation of simple and animals. Retinal W cells also have slowly conducting, poorly complex receptive field types. The finding that similar orien- myelinated axons (Cleland and Levick, 1974a,b)consistent with tation tuning selectivities can be generatedin A 1 and V 1 further the relatively long latencies to optic chiasm stimulation found increasesthe likelihood that suchtuning properties are generated both in the MGN (Sur et al., 1988) and in Al. intracortically and are relatively independent of specific visual input characteristics. We argue, therefore, that the generation Contribution of intracortical mechanisms:similarity of of thesecortical receptivefield organizations and tuning prop- rewired Al to normal VI erties is not dependent on input cell type (i.e., X, Y, or W) but Single cells in A 1 exhibit aspectsof receptive field organization, arises via intracortical circuitries activated by a minimum of orientation selectivity, and direction selectivity absent in visual visual input. 3662 Roe et al. * Visual Responses from Auditory Cortex of Rewired Ferrets

Do inputs specify intracortical circuitry? that visual inputs to A 1 simply activate circuits normally present A possible interpretation of these data is that visual input during in A 1. A 1 thus comes to exhibit visual receptive field properties development has actually instructed the establishment of spe- similar to those in normal Vl. In this view, the common func- cific cortical circuits in A 1, which thereby exhibits the functional tional transformations that sensory cortices perform would in- properties similar to those normally seen in Vl. In this view, clude tuning for orientation and direction as well as generation the temporal patterns of sensory activity unique to each sensory of specific receptive field organization. modality actively instruct the development of specific cortical Anatomically, cortical laminae in Al and Vl are character- circuits possibly from an initially naive cortical template. This ized by similar patterns of inputs, outputs, as well as laminar is an attractive hypothesis in view of the significant literature cell type composition (e.g., Jones, 1984; Winer, 1984a,b, 1985; on activity-dependent mechanisms during cortical develop- Mitani and Shimokouchi, 1985); similar intralaminar projec- ment, and has been recently discussed in the context of this tion patterns (e.g., Gilbert and Wiesel, 1979; Mitani et al., 1985); experimental preparation (Pallas, 1990; Sur et al., 1990). and the presence ofpatchy local horizontal connectivity patterns It is clear that the development of functional properties of (Matsubara and Phillips, 1988; e.g., Gilbert and Wiesel, 1989b). cells in primary visual cortex can be modilicd by changes in Physiologically, perhaps akin to direction sclcctivity in V I, many visual input. Studies using abnormal visual rearing conditions neurons in Al respond selectively to the direction and rate of (for review, see Sherman and Spear, 1982), blockade of retinal frequency modulation (Whitfield and Evans, 1965; Mendelson activity (Chapman et al., 1986; Stryker and Harris, 1986), and and Cynader, 1985). In addition, similar to cells in normal V 1, perturbation of postsynaptic activity (Kasamatsu and Pettigrew, the tuning of Al cells is shaped importantly by intracortical 1979; Reiter and Stryker, 1988; Bear et al., 1990) during de- inhibition. Analogous to findings in both the adult and devel- velopment have shown that changing the quality and quantity oping visual cortex (Sillito, 1975; Fregnac et al., 1988; Greuel ofactivity can greatly influence the establishment ofocular dom- et al., 1988; Hata et al., 1988) local application of GABA an- inance in visual cortex. Other cortical receptive field properties tagonists such as bicuculline and excitatory agents such as ACh, (such as orientation selectivity, direction selectivity, and recep- norepinephrine, NMDA, and glutamate results in dramatic tive field organization) can also be affected by altering the visual changes in receptive field tuning properties of cells in A 1 (Mtiller environment. For example, binocular lid suture and dark rearing and Scheich, 1988; Ashe et al., 1989; McKenna et al., 1989a; produce dramatic reductions in the number of cells with normal Metherate and Weinberger, 1989). In addition, the study of activity levels and normal orientation and direction tuning prop- context-dependent responses in both visual (Gilbert and Wiesel, erties (Wiesel and Hubel, 1965; Pettigrew, 1974; Imbert and 1989a; Fox et al., 1990) and auditory cortices (Espinosa and Buisseret, 1975). Rearing in striped environments of a single Gerstein, 1988; McKenna et al., 1989b; Shamma et al., in press) orientation results in a proportionally higher incidence of cells supports the role of lateral inhibitory/modulatory mechanisms with similar orientation (Blakemore and Cooper, 1970; see, in shaping receptive field characteristics. These anatomical and however, Movshon and Van Sluyters, 198 1). functional parallels suggest that sensory cortices may indeed In this context, the functional similarity ofvisually innervated utilize some common processing mechanisms. Thus, the pres- auditory cortex or somatosensory cortex (Frost and Metin, 1985; ence of orientation-selective cells in Al of rewired ferrets may Metin and Frost, 1989) to normal visual cortex may be inter- suggest that similar “orientation tuning” transformations occur preted as the first functional demonstration of cross-modal in- normally in A 1. duction of modality-specific cortical circuitry by peripheral in- If sensory cortices share some common processing circuits, put. If significant differences exist in the intracortical circuitry these circuits can be established regardless of input modality of normal V 1 and normal A 1, these findings would suggest that during development. Consistent with this view, certain receptive inputs do not simply influence, but actually instruct the estab- field properties such as orientation tuning and simple/complex lishment of cortical circuits. At the time that the lesions are receptive field organization are already present in the young, made in the ferret, cerebral cortex is still quite immature. In visually inexperienced visual cortex (Hubel and Wiesel, 1963; primary visual cortex, only the deeper cortical layers (layers 4, Albus and Wolf, 1984; Braastad and Heggelund, 1985). The fact 5, and 6) have migrated into place and geniculocortical afferents that “normal” visual receptive field properties are found in A 1 are still waiting in the subplate (McConnell, 1988; Peduzzi, of rewired ferrets suggests that the development of many cortical 1988; Jackson et al., 1989; cf. Shatz and Luskin, 1986). Primary receptive field properties is not dependent on cell class-specific auditory cortex is likely to be at a similar stage of development (X, Y, or W) patterns of sensory input, and develop, in fact, (see, e.g., Feng and Brugge, 1983; Myslivecek, 1983). Thus, in despite a relatively weak source of retinal input (of the W type) the rewired ferrets, we have replaced auditory input with visual and despite the different pattern of thalamocortical connectivity input at a time when cortical circuits have not yet been fully that characterizes the auditory pathway (Pallas et al., 1990). established. It is likely that, at this stage in development, the That is not to say that normal cortical circuits develop in the microcircuitry of Al can be regulated significantly by afferent total absence of sensory input; a minimal amount of sensory input, regardless of modality (see also Schlaggar and O’Leary, input may be required to trigger events that result in establish- 199 1). It is possible that similar lesions made at an earlier stage ment of these cortical circuits. of development would produce an even more dramatic change in organization of the auditory pathway, such as changing the Concluding remarks divergent nature of auditory thalamocortical connectivity to a In conclusion, our results demonstrate that sensory neocortex point-to-point connectivity (cf. Pallas, 1990; Sur et al., 1990). is not uniquely specified to process inputs of a single modality. Auditory cortex, following specific lesions during development, Do sensory cortices share common processing circuits? can accept and process visual input. The visual transformations An alternative interpretation of our data is that Al and Vl performed by auditory cortex appear very similar to those in normally share at least certain similar processing circuits, and normal visual cortex. Whether these functional similarities are The Journal of Neuroscience, September 1992, 72(9) 3663 due to innut-induced changes in intracortical circuitrv or to Frost DO, Metin C (1985) Induction of functional retinal projections intrinsic commonalities acrosssensory cortices remains to be to the somatosensory system. Nature 3 17: 162-l 64. Fukuda Y, Hsiao CF, Watanabe M, Ito H (1984) Morphological cor- determined. A view consistent with that in other developing relates of physiologically identified Y-, X-, and W-cells in cat retina. systemsis that basic intracortical connections would be estab- J Neurophysiol 52:999-l 0 13. lished indenendent of afferents. while innut activitv would -nlav _ Gilbert CD (1977) Laminar differences in receptive field properties of a role in the elaboration or fine-tuning of these connections. cells in cat‘primary visual cortex. J Physiol (Land) 268:39 l-421. Such modulation by input activity would include alterations in Gilbert CD, Wiesel TN (1979) Morphology and intracortical projections of functionally characterized neurones in the cat visual cortex. the weightsof preexisting synapsesin the cortex (e.g., Merzenich Nature 280:120-125. et al., 1984; Donoghue and Sanes, 1988; Gilbert and Wiesel, Gilbert CD, Wiesel TN (1989a) The influence of contextual patterns 1989a;Mioche and Singer, 1989; Fox, 1990; Jenkinset al., 1990; on orientation selectivity in Vl of the cat. Sot Neurosci Abstr 15: Kaas et al., 1990). Indeed, our experiments on the mapping of 324. Gilbert CD, Wiesel TN (1989b) Columnar specificity of intrinsic hor- the visual field in Al of rewired animals (Roe et al., 1990a) izontal and corticocortical connections in cat visual cortex. J Neurosci indicate that one way for the map to form is by alterations in 9~2432-2442. intracortical inhibition due to visual inputs (Sur et al., 1990). Grcucl M, Luhmann HJ, Singer W (1988) Pharmacological induction of use-dependent receptive held modifications in the visual cortex. Science 242~74-77. Hata Y, Tsumoto T, Sato H, Hagihara K, and Tamura H (1988) In- References hibition contributes to orientation selectivity in visual cortex of the Albus K, Wolf W (1984) Early post-natal development of neuronal cat. Nature 335:815-S 17. function in the kitten’s visual cortex: a laminar analysis. J Physiol Hubel DH, Wiesel TN (1962) Receptive fields, binocular interaction, (Lond) 348: 153-l 85. and functional architecture in the cat’s visual cortex. J Physiol (Lond) Andersen RA, Knight PL, Merzenich MM (1980) The thalamocortical 160:106-154. and corticothalamic connections of AI, AH, and the anterior auditory Hubel DH, Wiesel TN (1963) Receptive fields of cells in striate cortex field (AAF) in the cat: evidence for two largely segregated systems of of very young, visually inexperienced kittens. J Neurophysiol26:994- connections. J Comp Neural 194:663-70 1. 1002. Ashe JH, McKenna TM, Weinberger NM (1989) Cholinergic mod- Imbert M, Buisseret P (1975) Receptive field characteristics and plastic ulation of frequency receptive fields in auditory cortex: II. Frequency- properties of visual cortical cells in kittens reared with or without specific effects of anticholinesterases provide evidence for a modu- visual experience. Exp Brain Res 22:25-36. latory action of endogenous ACh. Synapse 4~44-54. Jackson CA, Peduzzi JD, Hickey TL (1989) Visual cortex development

Bear MF. Kleinschmidt A. Gu 0. Sinaer W (1990)~ I Disruntion of ., ” in the ferret. I. Genesis and migration of visual cortical neurons. J experience-dependent synaptic modifications in striate cortex by in- Neurosci 9: 1242-l 253. fusion of an NMDA receptor antagonist. J Neurosci 10:909-925. Jenkins WM, Merzenich MM, Ochs MT, Allard T, Guic-Robles E Blakemore C, Cooper GF (1970) Development of the brain depends (1990) Functional reorganization of primary somatosensory cortex on the visual environment. Nature 228:477-478. in adult owl monkeys after behaviorally controlled tactile stimulation. Braastad BO, Heggelund P (1985) Development of spatial receptive J Neurophysiol 63:82-104. field organization and orientation selectivity in kitten striate cortex. Jones EG (1984) Identification and classification of intrinsic circuit J Neurophysiol 53: 1158-l 178. elements in the neocortex. In: Dynamic aspects ofneocortical function Chapman B, Jacobson MD, Reiter HO, Stryker MP (1986) Ocular (Edelman G, Gall E, Cowan WM, eds), pp 7-40. New York: Wiley. dominance shift in kitten visual cortex caused by imbalance in retinal Kaas JH, Krubitzer LA, Chino YM, Langston AL, Polley EH, Blair N electrical activity. Nature 324: 154-l 56. (1990) Reorganization of retinotopiccortical maps in adult mammals Cleland BG, Levick WR (1974a) Brisk and sluggish concentrically after lesions of the retina. Science 248:229-23 1. organized ganglion cells in the cat’s retina. J Physiol (Lond) 240:42 l- Kasamatsu T, Pettigrew JD (1979) Preservation of binocularity after 456. monocular deprivation in the striate cortex of kittens treated with Cleland BG, Levick WR (1974b) Properties of rarely encountered 6-hydroxydopamine. J Comp Neurol 185: 139-l 62. types of ganglion cells in the cat’s retina and an overall classification. Kelly JB, Judge PW, Phillips DP (1986) Representation of the cochlea J Physiol (Lond) 240:457492. in primary auditory cortex of the ferret (Must& putorius). Hearing Donoghue JP, Sanes JN (1988) Organization of adult motor cortex Res 24:ll l-l 15. representation patterns following neonatal forelimb nerve injury in Law MI, Zahs KR, Stryker MP (1988) Organization of primary visual rats. J Neurosci 8:3221-3232. - cortex (area 17) in the ferret. J Comp Neurol 278: 157-l 80. Esguerra M, Garraghty PE, Russo GS, Sur M (1986) Lateral geniculate LeVay S, McConnell SK, Luskin MB (1987) Functional organization nucleus in normal and monocularly deprived ferrets: X- and Y-cells of primary visualcortex in the mink (Mustela v&on), and a compar- and cell body size. Sot Neurosci Abstr-12: 10. ison with the cat. J Comp Neurol 257:422-441. Espinosa IE, Gerstein GL (1988) Cortical auditory neuron interactions Leventhal AG, Zhou Y, Ault SJ, Thompson KG (1990) Cortical con- during. presentation of 3-tone sequences: effective connectivity. Brain tribution to the orientation sensitivity of relay cells in the cat lateral Res 450:39-50. geniculate nucleus. Sot Neurosci Abstr 16: 176. Eysel UT, Worgotter F, Paper HC (1987) Local cortical lesions abolish Malpeli JG, Schiller PH, Colby CL (1981) Response properties of lateral inhibition at direction selective cells in visual cortex. EXP Brain single cells in monkey striate cortex during reversible inactivation of Res 68:606-6 12. individual lateral geniculate laminae. J Neurophysio146: 1102-l 119. Felleman DJ, Van Essen DC (1987) Receptive field properties of neu- Matsubara JA, Phillips DP (1988) Intracortical connections and their rons in area V3 ofmacaque monkey extrastriate cortex. J Neurophysi- physiological correlates in the primary auditory cortex (Al) of the 01 57:889-920. cat. J Comp Neurol 268:3848. Feng JZ, Brugge JF (1983) Postnatal development of auditory callosal McConnell SK (1988) Fates of visual cortical neurons in the ferret connections in the kitten. J Comp Neurol 2 14:4 16-126. after isochronic and heterochronic transplantation. J Neurosci 8:945- Ferster D (1990a) X- and Y-mediated synaptic potentials in neurons 974. of areas 17 and 18 of cat visual cortex. Vis Neurosci 4: 115-l 33. McKenna TM, Ashe JH, Weinberger NM (1989a) Cholinergic mod- Ferster D (1990b) X- and Y-mediated current sources in areas 17 and ulation of frequency receptive fields in auditory cortex: I. Frequency- 18 of cat visual cortex. Vis Neurosci 4: 135-l 45. specific effects of muscarinicagonists. Synapse 4:30-43. Fox JM, Delbruck T, Gallant JL, Anderson CH, Van Essen DC (1990) McKenna TM, Weinberger NM, Diamond DM (1989b) Responses of Modulation of classical receptive field responses by moving texture single auditory cortical neurons to tone sequences. Brain Res 481: backgrounds in monkey striate cortex: spatial and temporal interac- 142-153. tions. Sot Neurosci Abstr 16: 1270. Mendelson JR, Cynader MS (1985) Sensitivity ofcat primary auditory Fregnac Y, Shulz D, Thorpe S, Bienenstock E (1988) A cellular ana- cortex (Al) neurons to the direction and rate offrequency modulation. logue of visual cortical plasticity. Nature 333:367-370. Brain Res 327:33 l-335. 3664 Roe et al. - Visual Responses from Auditory Cortex of Rewired Ferrets

Merzenich MM, Jenkins WM, Middlebrooks JC (1984) Observations thalamus and cortex: an experimentally induced visual pathway in and hypotheses on special organizational features of the central au- ferrets. PhD thesis, MIT. ditory nervous system. In: Dynamic aspects of neocortical function Roe AW, Garraghty PE, Sur M (1987) Retinotectal W-cell plasticity: (Edelman GM, Gall WE, Cowan WM, eds), pp 397424. New York: experimentally induced retinal projections to auditory thalamus in Neurosciences Research Foundation. ferrets. Sot Neurosci Abstr 13: 1023. Metherate R, Weinberger NM (1989) Acetylcholine produces stimu- Roe AW, Garraghty PE, Sur M (1989) The terminal arbors of single lus-specific receptive field alterations in cat auditory cortex. Brain Res on-center and off-center X and Y retinogeniculate axons within the 480~372-377. ferret’s lateral geniculate nucleus. J Comp Neurol 288:208-242. Metin C, Frost DO (1989) Visual responses of neurons in somatosen- Roe AW, Pallas SL, Hahm JO, Sur M (1990a) A map of visual space sory cortex of hamsters with experimentally induced retinal projec- induced into primary auditory cortex. Science 250:8 18-820. tions to somatosensory thalamus. Proc Nat1 Acad Sci USA 86:357- Roe AW, Pallas SL, Kwon YH, Sur M (1990b) Visually innervated 361. auditory cortex in ferrets generates visual response properties similar Middlebrooks JC, Zook JM (1983) Intrinsic organization of the cat’s to those in normal visual cortex. Sot Neurosci Abstr 16:984. medial geniculate body identified by projections to binaural response- Schiller PH, Finlay BL, Volman SF (1976a) Quantitative studies of specific bands in the primary auditory cortex. J Neurosci 3:203-224. single-cell properties in monkey striate cortex. I. Spatiotemporal or- Mioche L, Singer W (1989) Chronic recordings from single sites of gamzation,f rcccptive fields. J- Neurophysiol 39: 1288-l 3 19. kitten striate cortex during expcricnce-dependent modilications of Schiller PH. Finlav BL. Volman SF (1976b) , Ouantitative. studies of receptive-field properties. J Ncurophysiol 62:185-197. single-cell properties ‘in monkey striate cortex. II. Orientation spec- Mitani A, Shimokouchi M (1985) Neuronal connections in the pri- ificity and ocular dominance. J Neurophysiol 39: 1320-l 333. mary auditory cortex: an electrophysiological study in the cat. J Comp Schlaggar BL, O’Leary DDM (199 1) Potential of visual cortex to de- Neurol 235:417-429. velop an array of functional units unique to somatosensory cortex. Mitani A, Shimokouchi M, Itoh K, Nomura S, Kudo M, Mizuno N Science 252: 1556-l 560. (1985) Morphology and laminar organization of electrophysiolog- Schneider GE (1973) Early lesions of superior colliculus: factors af- ically identified neurons in the primary auditory cortex in the cat. J fecting the formation of abnormal retinal projections. Brain Behav Comp Neurol 235:430-447. Evol 8:73-109. Movshon JA, Van Sluyters RC (1981) Visual neural development. Shamma SA, Fleshman JW, Wiser PR (1992) Receptive field orga- Annu Rev Psycho1 321477-522. nization in ferret primary auditory cortex: spectral orientation col- Mtiller CM, Scheich H (1988) Contribution of GABAergic inhibition umns. J Neurophysiol, in press. to the response characteristics of auditow units in the avian forebrain. Shatz CJ, Luskin MB (1986) The relationship between the geniculo- J Neurophysiol 59:1673-1689. - cortical afferents and their cortical target cells during development of the cat’s primary visual cortex. J Neurosci 6:3655-3668. Mullikin WH. ,- Jones JP.-I Palmer LA (1984) Retentive field _nronerties _ and laminar distribution of X-like and Y-like simple cells in cat area Sherman SM, Spear PD (1982) Organization of visual pathways in 17. J Neurophysiol 52:350-37 1. normal and visually deprived cats. Physiol Rev 62:738-855. Myslivecek J (1983) Development of the auditory evoked responses Sillito AM (1975) The contribution of inhibitory mechanisms to the in the auditory cortex in mammals. In: Development of auditory and receptive field properties of neurons in the striate cortex of the cat. J vestibular systems (Romand R, ed), pp 167-209. New York: Aca- Physiol (Lond) 250:305-322. demic. Stanford LR (1987) W-cells in the cat retina: correlated morphological Orban GA, Kennedy H, Maes H (198 1) Response to movement of and physiological evidence for two distinct classes. J Neurophysiol neurons in areas 17 and 18 of the cat: velocity sensitivity. J Neuro- 57:2 18-244. physiol 45:1043-1058. Stryker MP, Harris WA (1986) Binocular impulse blockade prevents Pallas SL (1990) Cross-modal plasticity in sensory cortex. In: The the formation of ocular dominance columns in cat visual cortex. J neocortex: ontogeny and phylogeny (Finlay BL, Innocenti G, Scheich Neurosci 6:2 117-2 133. H, eds), pp 205-2 18. New York: Plenum. Sur M. Sherman SM (1982) Linear and nonlinear W-cells in C-laminae Pallas SL, Roe AW, Sur M (1988) Retinal projections induced into of the cat’s lateral geniculate nucleus. J Neurophysiol47:869-884. auditory thalamus in ferrets: changes in inputs and outputs of primary Sur M, Garraghty PE, Roe AW (1988) Experimentally induced pro- auditory cortex. Sot Neurosci Abstr 14:460. jections into auditory thalamus and cortex. Science 242:1437-1441. Pallas SL, Hahm JO, Sur M (1989) Retinal axon arbors in a novel Sur M, Pallas SL, Roe AW (1990) Cross-modal plasticity in cortical target: morphology of ganglion cell axons induced to arborize in the development: differentiation and specification of sensory neocortex. medial geniculate nucleus of ferrets. Sot Neurosci Abstr 15:495. Trends Neurosci 13:227-233. Pallas SL, Roe AW, Sur M (1990) Visual projections induced into the Tanaka K (1983) Cross-correlation analysis ofgeniculostriate neuronal auditory pathway offerrets. I. Novel inputs to primary auditory cortex relationships in cats. J Neurophysiol 49: 1303-l 3 18. (A 1) from the LP/pulvinar complex and the topography of the MGN- Vitek DJ, Schall JD, Leventhal AG (1985) Morphology, central pro- Al projection. J Comp Neurol 298:50-68. jections, and dendritic field orientation of retinal ganglion cells in the Palmer LA, Davis TL (1981) Receptive field structure in cat striate ferret. J Comp Neurol 241: l-l 1. cortex. J Neurophysiol46:260-276. Whitfield IC, Evans EF (1965) Responses of auditory cortical neurons Palmer LA, Rosenquist AC (1974) Visual receptive fields of single to stimuli of changing frequency. J Neurophysiol 28:655-672. striate cortical units projecting to the superior colliculus in the cat. Wiesel TN, Hubel DH (1965) Comparison of the effects of unilateral Brain Res 67:2742. and bilateral eye closure on cortical unit responses in kittens. J Neu- Peduzzi JD (1988) Genesis of GABA-immunoreactive neurons in the rophysiol 28:1029-1040. ferret visual cortex. J Neurosci 8:920-93 1. Wilson PD, Rowe MH, Stone J (1976) Properties of relay cells in the Pettigrew JD (1974) The effectof visual experience on the development cat’s lateral geniculate nucleus: a comparison of W-cells with X- and of stimulus specificity by kitten cortical neurones. J Physiol 237:49- Y-cells. J Neurophysiol 39: 1193-l 209. 14. Winer JA (1984a) Anatomy of layer IV in cat primary auditory cortex Price DJ, Morgan JE (1987) Spatial properties ofneurones in the lateral (Al). J Comp Neurol 224:535-567. geniculate nucleus of the pigmented ferret. Exp Brain Res 68:28-36. Winer JA (1984b) The non-pyramidal cells in layer III of cat primary Redies C, Diksic M, Rimi H (1990) Functional organization in the auditory cortex (Al). J Comp Neurol 229:512-530. ferret visual cortex: a double-label 2-deoxyglucose study. J Neurosci Winer JA (1985) Structure of layer II in cat primary auditory cortex 10:2791-2803. (Al). J Comp Neurol 238310-37. Reiter HO, Stryker MP (1988) Neural plasticity without action po- Zahs KR. Strvker MP (1985) The nroiection of the visual field onto tentials: less active inputs become dominant when kitten visual cor- the lateral geniculate nucleus of the ferret. J Comp Neurol 24 1:2 1O- tical cells are pharmacologically inhibited. Proc Nat1 Acad Sci USA 224. 8513623-3627. Zahs KR, Stryker MP (1988) Segregation of ON and OFF aflerents to Roe AW (199 1) Functional transformations of visual input by auditory ferret visual cortex. J Neurophysiol 59:1410-1429.