106

J. Phyiiol. (1962), 160, pp. 106-154
With 2 plate and 20 text-ftgutre8
Printed in Gret Britain

RECEPTIVE FIELDS, BINOCULAR INTERACTION
AND FUNCTIONAL ARCHITECTURE IN
THE CAT'S VISUAL CORTEX

BY D. H. HUBEL AD T. N. WIESEL
From the Neurophysiolojy Laboratory, Department of Pharmacology
Harvard Aledical School, Boston, Massachusetts, U.S.A.

(Received 31 July 1961)

What chiefly distinguishes cerebral cortex from other parts of the central nervous system is the great diversity of its cell types and interconnexions. It would be astonishing if such a structure did not profoundly modify the response patterns of fibres coming into it. In the cat's visual cortex, the receptive field arrangements of single cells suggest that there is indeed a degree of complexity far exceeding anything yet seen at lower levels in the visual system.
In a previous paper we described receptive fields of single cortical cells, observing responses to spots of light shone on one or both retinas (Hubel
& Wiesel, 1959). In the present work this method is used to examine receptive fields of a more complex type (Part I) and to make additional observations on binocular interaction (Part II).
This approach is necessary in order to understand the behaviour of individual cells, but it fails to deal with the problem of the relationship of one cell to its neighbours. In the past, the technique of recording evoked slow waves has been used with great success in studies of functional anatomy. It was employed by Talbot & Marshall (1941) and by Thompson, Woolsey & Talbot (1950) for mapping out the visual cortex in the rabbit, cat, and monkey. Daniel & Whitteiidge (1959) have recently extended this work in the primate. Most of our present knowledge of retinotopic projections, binocular overlap, and the second visual area is based on these investigations. Yet the method of evoked potentials is valuable mainly for detecting behaviour common to large populations of neighbouring cells; it cannot differentiate functionally between areas of cortex smaller than about 1 mm2. To overcome this difficulty a method has in recent years been developed for studying cells separately or in small groups during long micro-electrode penetrations through nervous tissue.
Responses are correlated with cell location by reconstructing the electrode tracks from histological material. These techniques have been applied to
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

107
CAT VISUAL CORTEX the somatic sensory cortex of the cat and monkey in a remarkable series of studies by Mountcastle (1957) and Powell & Mountcastle (1959). Their results show that the approach is a powerful one, capable of revealing systems of organization not hinted at by the known morphology. In
Part III of the present paper we use this method in studying the functional architecture of the visual cortex. It helped us attempt to explain on anatomical grounds how cortical receptive fields are built up.
METHODS
Recordings were made from forty acutely prepared cats, anaesthetized with thiopental sodium, and maintained in light sleep with additional doses by observing the electrocorticogram. Animals were paralysed with succinylcholine to stabilize the eyes. Pupils were dilated with atropine. Details of stimulating and recording methods are given in previous papers (Hubel, 1959; Hubel & Wiesel, 1959, 1960). The animal faced a wide tangent screen at a distance of 1-5 m, and various patterns of white light were shone on the screen by a tungsten-filament projector. All recordings were made in the lightadapted state. Background illumination varied from - 1-0 to + 1.0 log1o cd/M2. Stimuli were from 0-2 to 2-0 log. units brighter than the background. For each cell receptive fields were mapped out separately for the two eyes on sheets of paper, and these were kept as permanent records.
Points on the screen corresponding to the area centralis and the optic disk of the two eyes were determined by a projection method (Hubel & Wiesel, 1960). The position of each receptive field was measured with respect to these points. Because of the muscle relaxant the eyes usually diverged slightly, so that points corresponding to the two centres of gaze were not necessarily superimposed. In stimulating the two eyes simultaneously it was therefore often necessary to use two spots placed in corresponding parts of the two visual fields. Moreover, at times the two eyes were slightly rotated in an inward direction in the plane of their equators. This rotation was estimated by (1) photographing the cat before and during the experiment, and comparing the angles of inclination of the slit-shaped pupils, or (2) by noting the inclination to the horizontal of a line joining the area centralis with the optic disk, which in the normal position of the eye was estimated, by the first method, to average about 250. The combined inward rotations of the two eyes seldom exceeded 100.
Since the receptive fields in this study were usually centrally rather than peripherally placed on the retina, the rotations did not lead to any appreciable linear displacement. Angular displacements of receptive fields occasionally required correction, as they led to an apparent difference in the orientation of the two receptive-field axes of a binocularly driven unit.
The direction and magnitude of this difference were always consistent with the estimated inward rotation of the two eyes. Moreover, in a given experiment the difference was constant, even though the axis orientation varied from cell to cell.
The diagram of Text-fig. 1 shows the points of entry into the cortex of all 45 microelectrode penetrations. Most electrode tracks went no deeper than 3 or 4 mm, so that explorations were mainly limited to the apical segments of the lateral and post-lateral gyri
(LG and PLG) and a few millimetres down along the adjoining medial and lateral folds. The extent of the territory covered is indicated roughly by Text-figs. 13-15. Although the lateral boundary of the striate cortex is not always sharply defined in Nissl-stained or myelinstained material, most penetrations were well within the region generally accepted as
'striate' (O'Leary, 1941). Most penetrations were made from the cortical region receiving projections from in or near the area centralis; this cortical region is shown in Text-fig. 1 as the area between the interrupted lines.

Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

108

D. H. HUBEL AND T. N. WIESEL

Tungsten micro-electrodes were advanced by a hydraulic micro-electrode positioner
(Hubel, 1957, 1959). In searching for single cortical units the retina was continually stimulated with stationary and moving forms while the electrode was advanced. The unresolved background activity (see p. 129) served as a guide for determining the optimum stimulus.
This procedure increased the number of cells observed in a penetration, since the sampling was not limited to spontaneously active units.
In each penetration electrolytic lesions were made at one or more points. When only one lesion was made, it was generally at the end of an electrode track. Brains were fixed in 10 % formalin, embedded in celloidin, sectioned at 20 u, and stained with cresyl violet. Lesions were 50-100 I in diameter, which was small enough to indicate the position of the electrode tip to the nearest cortical layer. The positions of other units encountered in a cortical penetration were determined by calculating the distance back from the lesion along the track,

Text-fig. 1. Diagram of dorsal aspect of cat's brain, to show entry points of 45 micro-electrode penetrations. The penetrations between the interrupted lines are those in which cells had their receptive fields in or near area centralis. LG, lateral gyrus; PLG, post-lateral gyrus. Scale, 1 cm.

using depth readings corresponding to the unit and the lesion. A correction was made for brain shrinkage, which was estimated by comparing the distance between two lesions, measured under the microscope, with the distance calculated from depths at which the two lesions were made. From brain to brain this shrinkage was not constant, so that it was not possible to apply an average correction for shrinkage to all brains. For tracks marked by only one lesion it was assumed that the first unit activity was recorded at the boundary of the first and second layers; any error resulting from this was probably small, since in a number of penetrations a lesion was made at the point where the first units were encountered, and these were in the lower first or the upper second layers, or else at the very boundary. The absence of cell-body records and unresolved background activity as the electrode passed through subcortical white matter (see Text-fig. 13 and P1. 1) was also helpful in confirming the accuracy of the track reconstructions.

Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

CAT VISUAL CORTEX

109

PART I
ORGANIZATION OF RECEPTIVE FIELDS IN CAT'S
VISUAL CORTEX: PROPERTIES OF 'SIMPLE'
AND 'COMPLEX' FIELDS
The receptive field of a cell in the visual system may be defined as the region of retina (or visual field) over which one can influence the firing of that cell. In the cat's retina one can distinguish two types of ganglion cells, those with 'on'-centre receptive fields and those with 'off'-centre fields (Kuffler, 1953). The lateral geniculate body also has cells of these two types; so far no others have been found (Hubel & Wiesel, 1961). In contrast, the visual cortex contains a large number of functionally different cell types; yet with the exception of afferent fibres from the lateral geniculate body we have found no units with concentric 'on'-centre or
'off'-centre fields.
When stimulated with stationary or moving patterns of light, cells in the visual cortex gave responses that could be interpreted in terms of the arrangements of excitatory and inhibitory regions in their receptive fields
(Hubel & Wiesel, 1959). Not all cells behaved so simply, however; some responded in a complex manner which bore little obvious relationship to the receptive fields mapped with small spots. It has become increasingly apparent to us that cortical cells differ in the complexity of their receptive fields. The great majority of fields seem to fall naturally into two groups, which we have termed 'simple' and 'complex'. Although the fields to be described represent the commonest subtypes of these groups, new varieties are continually appearing, and it is unlikely that the ones we have listed give anything like a complete picture of the striate cortex.
We have therefore avoided a rigid system of classification, and have designated receptive fields by letters or numbers only for convenience in referring to the figures. We shall concentrate especially on features common to simple fields and on those common to complex fields, emphasizing differences between the two groups, and also between cortical fields and lateral geniculate fields.
RESULTS

Simple receptive fields
The receptive fields of 233 of the 303 cortical cells in the present series were classified as 'simple'. Like retinal ganglion and geniculate cells,

cortical cells with simple fields possessed distinct excitatory and inhibitory subdivisions. Illumination of part or all of an excitatory region increased the maintained firing of the cell, whereas a light shone in the
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

D. H. HUBEL AND T. N. WIESEL inhibitory region suppressed the firing and evoked a discharge at 'off'. A large spot confined to either area produced a greater change in rate of firing than a small spot, indicating summation within either region. On the other hand, the two types of region within a receptive field were mutually antagonistic. This was most forcefully shown by the absence or near absence of a response to simultaneous illumination of both regions, for example, with diffuse light. From the arrangement of excitatory and inhibitory regions it was usually possible to predict in a qualitative way the responses to any shape of stimulus, stationary or moving. Spots having the approximate shape of one or other region were the most effective stationary stimuli; smaler spots failed to take full advantage of summation within a region, while larger ones were likely to invade opposing regions, so reducing the response. To summarize: these fields were termed 'simple' because like retinal and geniculate fields (1) they were subdivided into distinct excitatory and inhibitory regions; (2) there was summation within the separate excitatory and inhibitory parts; (3) there was antagonism between excitatory and inhibitory regions; and (4) it was possible to predict responses to stationary or moving spots of various shapes from a map of the excitatory and inhibitory areas.
While simple cortical receptive fields were similar to those of retinal ganglion cells and geniculate cells in possessing excitatory and inhibitory subdivisions, they differed profoundly in the spatial arrangements of these regions. The receptive fields of all retinal ganglion and geniculate cells had one or other of the concentric forms shown in Text-fig. 2 A, B. (Excitatory areas are indicated by crosses, inhibitory areas by triangles.) In contrast, simple cortical fields all had a side-to-side arrangement of excitatory and inhibitory areas with separation of the areas by parallel straight-line boundaries rather than circular ones. There were several varieties of fields, differing in the number of subdivisions and the relative area occupied by each subdivision. The commonest arrangements are illustrated in Text-fig.
2C-G: Table 1 gives the number of cels observed in each category. The departure of these fields from circular symmetry introduces a new variable, namely, the orientation of the boundaries separating the field subdivisions.
This orientation is a characteristic of each cortical cell, and may be vertical, horizontal, or oblique. There was no indication that any one orientation was more common than the others. We shall use the term receptive-field axis to indicate a line through the centre of a field, parallel to the boundaries separating excitatory and inhibitory regions. The axis orientation wiU then refer to the orientation of these boundaries, either on the retina or in the visual field. Axes are shown in Text-fig. 2 by continuous lines.
Two common types of fields, shown in Text-fig. 2C, D, each consisted of a narrow elongated area, excitatory or inhibitory, flanked on either side
110

Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

III
CAT VISUAL CORTEX by two regions of the opposite type. In these fields the two flanking regions were symmetrical, i.e. they were about equal in area and the responses obtained from them were of about the same magnitude. In addition there were fields with long narrow centres (excitatory or inhibitory) and asymmetrical flanks. An example of an asymmetrical field with an inhibitory centre is shown in Text-fig. 2E. The most effective stationary stimulus for all of these celLs was a long narrow rectangle ('slit') of light just large
A

c

D

4 D*

X4
A

4

FG

E

t

k~~~~
A~~~~~~~~~~~~~~A

4 A4-

Text-fig. 2. Common arrangements of lateral geniculate and cortical receptive fields. A. 'On'-centre geniculate receptive field. B. 'Off'-centre geniculate receptive field. 0-G. Various arrangements of simple cortical receptive fields. x, areas giving excitatory responses ('on' responses); A, areas giving inhibitory responses ('off' responses). Receptive-field axes are shown by continuous lines through field centres; in the figure these are all oblique, but each arrangement occurs in all orientations.

enough to cover the central region without invading either flank. For maximum centre response the orientation of the slit was critical; changing the orientation by more than 5l10 was usually enough to reduce a response greatly or even abolish it. Illuminating both flanks usually evoked a strong response. If a slit having the same size as the receptive-field centre was shone in either flanking area it evoked only a weak response, since it covered only part of one flank. Diffuse light was ineffective, or at most evoked only a very weak response, indicating that the excitatory and inhibitory parts of the receptive field were very nearly balanced.
In these fields the equivalent but opposite-type regions occupied retinal
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

112
D. H. HUBEL AND T. N. WIESEL areas that were far from equal; the centre portion was small and concentrated whereas the flanks were widely dispersed. A similar inequality was found in fields of type F, Text-fig. 2, but here the excitatory flanks were elongated and concentrated, while the centre was relatively large and diffuse. The optimum response was evoked by simultaneously illuminating the two flanks with two parallel slits (see Hubel & Wiesel, 1959, Fig. 9).
Some cells had fields in which only two regions were discernible, arranged side by side as in Text-fig. 2 G. For these cells the most efficient stationary stimulus consisted of two areas of differing brightness placed so that the line separating them fell exactly over the boundary between the excitatory and inhibitory parts of the field. This type of stimulus was termed an
'edge'. An 'on' or an 'off' response was evoked depending on whether the bright part of the stimulus fell over the excitatory or the inhibitory region.
A slight change in position or orientation of the line separating the light from the dark area was usually enough to reduce greatly the effectiveness of the stimulus.
Moving stimuli were very effective, probably because of the synergistic effects of leaving an inhibitory area and simultaneously entering an excitatory area (Hubel & Wiesel, 1959). The optimum stimulus could usually be predicted from the distribution of excitatory and inhibitory regions of the receptive field. With moving stimuli, just as with stationary, the orientation was critical. In contrast, a slit or edge moved across the circularly symmetric field of a geniculate cell gave (as one would expect) roughly the same response regardless of the stimulus orientation. The responses evoked when an optimally oriented slit crossed back and forth over a cortical receptive field were often roughly equal for the two directions of crossing. This was true of fields like those shown in Text-fig. 2C,
D and F. For many cells, however, the responses to two diametrically opposite movements were different, and some only responded to one of the two movements. The inequalities could usually be accounted for by an asymmetry in flanking regions, of the type shown in Text-fig. 2E (see also
Hubel & Wiesel, 1959, Fig. 7). In fields that had only two discernible regions arranged side by side (Text-fig. 2 G), the difference in the responses to a moving slit or edge was especially pronounced.
Optimum rates of movement varied from one cell to another. On several occasions two cells were recorded together, one of which responded only to a slow-moving stimulus (1°/sec or lower) the other to a rapid one
(10°/sec or more). For cells with fields of type F, Text-fig. 2, the time elapsing between the two discharges to a moving stimulus was a measure of the rate of movement (see Hubel & Wiesel, 1959, Fig. 5).
If responses to movement were predictable from arrangements of excitatory and inhibitory regions, the reverse was to some extent also true.

Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

CAT VISUAL CORTEX
113
The axis orientation of a field, for example, was given by the most effective orientation of a moving slit or edge. If an optimally oriented slit produced a brief discharge on crossing from one region to another, one could predict that the first region was inhibitory and the second excitatory.
Brief responses to crossing a very confined region were characteristic of cells with simple cortical fields, whereas the complex cells to be described below gave sustained responses to movement over much wider areas.
TABLE 1. Simple cortical fields

Text-fig.

(a) Narrow concentrated centres
(i) Symmetrical flanks
Excitatory centres
Inhibitory centres
(ii) Asynmetrical flanks
Excitatory centres
Inhibitory centres
(b) Large centres; concentrated flanks
(c) One excitatory region and one inhibitory
(d) Uncategorized
Total number of simple fields

No. of cells

20

23
17

2D
2E
2F
2G

28
10
21
17
117
233

Movement was used extensively as a stimulus in experiments in which the main object was to determine axis orientation and ocular dominance for a large number of cells in a single penetration, and when it was not practical, because of time limitations, to map out every field completely.
Because movement was generally a very powerful stimulus, it was also used in studying cells that gave little or no response to stationary patterns.
In all, 117 of the 233 simple cells were studied mainly by moving stimuli.
In Table 1 these have been kept separate from the other groups since the distribution of their excitatory and inhibitory regions is not known with the same degree of certainty. It is also possible that with further study, some of these fields would have revealed complex properties.

Complex receptive fields
Intermixed with cells having simple fields, and present in most penetrations of the striate cortex, were cells with far more intricate and elaborate properties. The receptive fields of these cells were termed 'complex'.
Unlike cells with simple fields, these responded to variously-shaped stationary or moving forms in a way that could not be predicted from maps made with small circular spots. Often such maps could not be made, since small round spots were either ineffective or evoked only mixed ('on-off') responses throughout the receptive field. VVhen separate 'on' and 'off' regions could be discerned, the principles of summation and mutual antagonism, so helpful in interpreting simple fields, did not generally hold.
Nevertheless, there were some important features common to the two
8

PhYsiOl. 160
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

114
D. H. HUBEL AND T. N. WIESEL types of cells. In the following examples, four types of complex fields will be illustrated. The numbers observed of each type are given in Table 2.

(a)
(b)
(c)
(d)

TABLE 2. Complex cortical receptive fields
Text-fig.
Activated by slit-non-uniform field
3
4
Activated by slit-uniform field
Activated by edge
5-6
Activated by dark bar
7-8
Total number of complex fields

No. of cells
11
39
14
6
70

The cell of Text-fig. 3 failed to respond to round spots of light, whether small or large. By trial and error with many shapes of stimulus it was discovered that the cell's firing could be influenced by a horizontally oriented slit J' wide and 30 long. Provided the slit was horizontal its exact
A

BS

l

D

l

I'l f
I

ll

II. w |

llm

EIi

F

.f

LI

T

fl

G
Text-fig. 3. Responses of a cell with a complex receptive field to stimulation of the left (contralateral) eye. Receptive field located in area centralis. The diagrams to the left of each record indicate the position of a horizontal rectangular light stimulus with respect to the receptive field, marked by a cross. In each record the upper line indicates when the stimulus is on. A-E, stimulus i x 30, F-G, stimulus
1j x 3° (40 is equivalent to 1 mm on the cat retina). For background illumination and stimulus intensity see Methods. Cell was activated in the same way from right eye, but less vigorously (ocalar-dominance group 2, see Part II). An electrolytic lesion made while recording from this cell was found near the border of layers 5 and 6, in the apical segment of the post-lateral gyrus. Positive deflexions upward; duration of each stimulus 1 sec.

Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

115
CAT VISUAL CORTEX positioning within the 3°-diameter receptive field was not critical. When it was shone anywhere above the centre of the receptive field (the horizontal line of Text-fig. 3) an 'off' response was obtained; 'on' responses were evoked throughout the lower half. In an intermediate position (Text-fig.
3 C) the cell responded at both 'on' and 'off'. From experience with simpler receptive fields one might have expected wider slits to give increasingly better responses owing to summation within the upper or lower part of the field, and that illumination of either half by itself might be the most effective stimulus of all. The result was just the opposite: responses fell off rapidly as the stimulus was widened beyond about J°, and large rectangles covering the entire lower or upper halves of the receptive field were quite ineffective (Text-fig. 3F, C). On the other hand, summation could easily be demonstrated in a horizontal direction, since a slit lo wide but extending only across part of the field was less effective than a longer one covering the entire width. One might also have expected the orientation of the slit to be unimportant as long as the stimulus was wholly confined to the region above the horizontal line or the region below. On the contrary, the orientation was critical, since a tilt of even a few degrees from the horizontal markedly reduced the response, even though the slit did not cross the boundary separating the upper and lower halves of the field.
In preferring a slit specific in width and orientation this cell resembled certain cells with simple fields. When stimulated in the upper part of its field it behaved in many respects like cells with 'off'-centre fields of type
D, Text-fig. 2; in the lower part it responded like 'on'-centre fields of
Text-fig. 2C. But for this cell the strict requirements for shape and orientation of the stimulus were in marked contrast to the relatively large leeway of the stimulus in its ordinate position on the retina. Cells with simple fields, on the other hand, showed very little latitude in the positioning of an optimally oriented stimulus.
The upper part of this receptive field may be considered inhibitory and the lower part excitatory, even though in either area summation only occurred in a horizontal direction. Such subdivisions were occasionally found in complex fields, but more often the fields were uniform in this respect. This was true for the other complex fields to be described in this section.
Responses of a second complex unit are shown in Text-fig. 4. In many ways the receptive field of this cell was similar to the one just described.
A slit was the most potent stimulus, and the most effective width was again K0. Once more the orientation was an important stimulus variable, since the slit was effective anywhere in the field as long as it was placed in a 10 o'clock-4 o'clock orientation (Text-fig. 4A-D). A change in orientation of more than 5-10' in either direction produced a marked
8-2
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

~~ A

116
D. H. HIUBEL AND T. N. WIESEL reduction in the response (Text-fig. 4E-G). As usual, diffuse light had no influence on the firing. This cell responded especially well if the slit, oriented as in A-D, was moved steadily across the receptive field.
Sustained discharges were evoked over the entire length of the field. The optimum rate of movement was about 1°/sec. If movement was interrupted the discharge stopped, and when it was resumed the firing recommenced. Continuous firing could be maintained indefinitely by small sideA

I

W ii C

'Xs4

i ~ ul TI~~ Iii fl oil11 Al 1F I II T11 'I F

DX

E

I

't

tk
Affiih]

_TTT

I IIIIII_111

-1OWF FkLFt

F

i

;

G
H

~V

1 r uIIll-{l l iiiu1 i ~ r

JI I I 111.

H
Text-fig. 4. Responses of a cell with a complex field to stimulation of the left
(contralateral) eye with a slit j x 2ij. Receptive field was in the area centralis and was about 2 x 30 in size. A-D, jO wide slit oriented parallel to receptive field axis. E-G, slit oriented at 45 and 90 to receptive-field axis. H, slit oriented as in
A-D, is on throughout the record and is moved rapidly from side to side where indicated by upper beam. Responses from left eye slightly more marked than those from right (Group 3, see Part II). Time 1 sec.

to-side movements of a stimulus within the receptive field (Text-fig. 4H).
The pattern of firing was one characteristic of many complex cells, especially those responding well to moving stimuli. It consisted of a series of short high-frequency repetitive discharges each containing 5-10 spikes. The
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

117
CAT VISUAL CORTEX bursts occurred at irregular intervals, at frequencies up to about 20/sec.
For this cell, movement of an optimally oriented slit was about equally effective in either of the two opposite directions. This was not true of all complex units, as will be seen in some of the examples given below.
Like the cell of Text-fig. 3 this cell may be thought of as having a counterpart in simple fields of the type shown in Text-fig. 2 C-E. It shares with these simpler fields the attribute of responding well to properly oriented slit stimuli. Once more the distinction lies in the permissible variation in position of the optimally oriented stimulus. The variation is small (relative to the size of the receptive field) in the simple fields, large in the complex. Though resembling the cell of Text-fig. 3 in requiring a slit for a stimulus, this cell differed in that its responses to a properly oriented slit were mixed ('on-off') in type. This was not unusual for cells with complex fields. In contrast, cortical cells with simple fields, like retinal ganglion cells and lateral geniculate cells, responded to optimum restricted stimuli either with excitatory ('on') responses or inhibitory
('off') responses. When a stimulus covered opposing regions, the effects normally tended to cancel, though sometimes mixed discharges were obtained, the 'on' and 'off' components both being weak. For these simpler fields 'on-off' responses were thus an indication that the stimulus was not optimum. Yet some cells with complex fields responded with mixed discharges even to the most effective stationary stimuli we could find.
Among the stimuli tried were curved objects, dark stripes, and still more complicated patterns, as well as monochromatic spots and slits.
A third type of complex field is illustrated in Text-figs. 5 and 6. There were no responses to small circular spots or to slits, but an edge was very effective if oriented vertically. Excitatory or inhibitory responses were produced depending on whether the brighter area was to the left or the right (Text-fig. 5A, E). So far, these are just the responses one would expect from a cell with a vertically oriented simple field of the type shown in Text-fig. 2 G. In such a field the stimulus placement for optimum response is generally very critical. On the contrary, the complex unit responded to vertical edges over an unusually large region about 16° in length (Text-fig. 6). 'On' responses were obtained with light to the left
(A-D), and 'off' responses with light to the right (E-H), regardless of the position of the line separating light from darkness. When the entire receptive field was illuminated diffusely (I) no response was evoked. As with all complex fields, we are unable to account for these responses by any simple spatial arrangement of excitatory and inhibitory regions.
Like the complex units already described, this cell was apparently more concerned with the orientation of a stimulus than with its exact position in the receptive field. It differed in responding well to edges but poorly or
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

118
D. H. HUBEL AND T. N. WIESEL not at all to slits, whether narrow or wide. It is interesting in this connexion that exchanging an edge for its mirror equivalent reversed the response, i.e. replaced an excitatory response by an inhibitory and vice versa. The ineffectiveness of a slit might therefore be explained by supposing that the opposite effects of its two edges tended to cancel each other.
A

A~~~~~~L

io um

- --

_

G

__

_

_

__---_

_

__-

Text-fig. 5. Responses of a cell with a large (5 x 160) complex receptive field to an edge projected on the ipsilateral retina so as to cross the receptive field in various directions. (The screen is illuminated by a diffuse background light, at 0-0 log10 cd/in2. At the time of stimulus, shown by upper line of each record, half the screen, to one side of the variable boundary, is illuminated at 1.0 log,0 cd/in2,while the other half is kept constant.) A, vertical edge with light area to left, darker area to right. B-H, various other orientations of edge. Position of receptive field 200 below and to the left of the area centralis. Responses from ipsilateral eye stronger than those from contralateral eye (group 5, see Part II). Time 1 sec.

As shown in Text-fig. 6, the responses of the cell to a given vertical edge were consistent in type, being either 'on' or 'off' for all positions of the edge within the receptive field. In being uniform in its response-type it resembled the cell of Text-fig. 4. A few other cells of the same general category showed a similar preference for edges, but lacked this uniformity.
Their receptive fields resembled the field of Text-fig. 3, in that a given edge evoked responses of one type over half the field, and the opposite type over
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

119
CAT VISUAL CORTEX the other half. These fields were divided into two halves by a line parallel to the receptive-field axis: an edge oriented parallel to the axis gave 'on' responses throughout one of the halves and 'off' responses through the other. In either half, replacing the edge by its mirror image reversed the response-type. Even cells, which were uniform in their response-types, like those in Text-fig. 4-6, varied to some extent in the magnitude of their responses, depending on the position of the stimulus. Moreover, as with most cortical cells, there was some variation in responses to identical stimuli. S_ S | z

B

I

_

_

I

sli

e--

IIU

IF

I

E

,

I

F

-

Text-fig. 6. Same cell as in Text-fig. 5. A-H, responses to a vertical edge in various parts of the receptive field: A-D, brighter light to the left; E-H, brighter light to the right; I, large rectangle, 10 x 20°, covering entire receptive field. Time,
1 sec.

A final example is given to illustrate the wide range of variation in the organization of complex receptive fields. The cell of Text-figs. 7 and 8 was not strongly influenced by any form projected upon the screen; it gave only weak, unsustained 'on' responses to a dark horizontal rectangle against a light background, and to other forms it was unresponsive. A strong discharge was evoked, however, if a black rectangular object (for example, a piece of black tape) was placed against the brightly illuminated screen.
The receptive field of the cell was about 5 x 5°, and the most effective stimulus width was about 1'. Vigorous firing was obtained regardless of the position of the rectangle, as long as it was horizontal and within the receptive field. If it was tipped more than 100 in either direction no discharge was evoked (Text-fig. 7D, E). We have recorded several complex fields which resembled this one in that they responded best to black rectangles against a bright background. Presumably it isimportant to
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

D. H. HUBEL AND T. N. WIESEL
120
have good contrast between the narrow black rectangle and the background; this is technically difficult with a projector because of scattered light. Slow downward movement of the dark rectangle evoked a strong discharge throughout the entire 50 of the receptive field (Text-fig. 8A). If the movement was halted the cell continued to fire, but less vigorously.
A

1u1

11

u1 1

iI

11

------

B iiii ll l

l

,i.1 I,, I11,.11

c
D

- ---

E

0000---

- - - W-000

Text-fig. 7. Cell activated only by left (contralateral) eye over a field approximately 5 x 50, situated 100 above and to the left of the area centralis. The cell responded best to a black horizontal rectangle, J x 60, placed anywhere in the receptive field (A-C). Tilting the stimulus rendered it ineffective (D-E). The black bar was introduced against a light background during periods of 1 sec, indicated by the upper line in each record. Luminance of white background, 1 0 log10 cd/M2; luminance of black part, 0-0 log10 cd/M2. A lesion, made while recording from the cell, was found in layer 2 of apical segment of post-lateral gyrus.

Upward movement gave only weak, inconsistent responses, and left-right movement (Text-fig. 8B) gave no responses. Discharges of highest frequency were evoked by relatively slow rates of downward movement
(about 5-10 sec to cross the entire field); rapid movement in either direction gave only very weak responses.
Despite its unusual features this cell exhibited several properties typical of complex units, particularly the lack of summation (except in a horizontal sense), and the wide area over which the dark bar was effective.
One may think of the field as having a counterpart in simple fields of type
D, Text-fig. 2. In such fields a dark bar would evoke discharges, but only if it fell within the inhibitory region. Moreover, downward movement of
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

121
CAT VISUAL CORTEX the bar would also evoke brisker discharges than upward, provided the upper flanking region were stronger than the lower one.
In describing simple fields it has already been noted that moving stimuli were often more effective than stationary ones. This was also true of cells with complex fields. Depending on the cell, slits, edges, or dark bars were most effective. As with simple fields, orientation of a stimulus was always critical, responses varied with rate of movement, and directional asymmetries of the type seen in Text-fig. 8 were common. Onlv once have we seen activation of a cell for one direction of movement and suppression of

i_+

A

+Bullli
I II1ta+

ll

iii iwp IIU mll
WU
-- l- 11110XWli I

__ l
I

11111

a 11 -

B.ii

T-

-11
I

--- --I
I-

ItIsIIII1
Text-fig. 8. Same cell as in Text-fig. 7. Movement of black rectangle j x 60 back and forth across the receptive field: A, horizontally oriented (parallel to receptivefield axis); B, vertically oriented. Time required to move across the field, 5 sec.
Time, 1 sec.

maintained firing for the opposite direction. In their responses to movement, cells with complex fields differed from their simple counterparts chiefly in responding with sustained firing over substantial regions, usually the entire receptive field, instead of over a very narrow boundary separating excitatory and inhibitory regions.

Receptive-field dimensions
Over-all field dimensions were measured for 119 cells. A cell was included only if its field was mapped completely, and if it was situated in the area of central vision (see p. 135). Fields varied greatly in size from one cell to the next, even for cells recorded in a single penetration (see Text-fig. 15). In
Text-fig. 9 the distribution of cells according to field area is given separately for simple and complex fields. The histogram illustrates the variation in size, and shows that on the average complex fields were larger than simple ones.
Widths of the narrow subdivisions of simple fields (the centres of types
C, D and E or the flanks of type F, Text-fig. 2) also varied greatly: the smallest were 10-15 minutes of arc, which is roughly the diameter of the smalest field centres we have found for geniculate cells. For some cels
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

122
D. H. HUBEL AND T. N. WIESEL with complex fields the widths of the most effective slits or dark bars were also of this order, indicating that despite the greater overall field size these cells were able to convey detailed information. We wish to emphasize that in both geniculate and cortex the field dimensions tend to increase with distance from the area centralis, and that they differ even for a given location in the retina. It is consequently not possible to compare field sizes in the geniculate and cortex unless these variations are taken into account. This may explain the discrepancy between our results and the findings of Baumgartner (see Jung, 1960), that 'field centres' in the cortex are one half the size of those in the lateral geniculate body.
30

,,20
0

E

z

0-25 0.5

1

2

4

8

16

32

Area of receptive fields in (degrees of arc)2
Text-fig. 9. Distribution of 119 cells in the visual cortex with respect to the approximate area of their receptive fields. White columns indicate cells with simple receptive fields; shaded columns, cells with complex fields. Abscissa: area of receptive fields. Ordinate: number of celLs.

Responsiveness of cortical cells
Simple and complex fields together account for all of the cells we have recorded in the visual cortex. We have not observed cells with concentric fields. Except for clearly injured cells (showing extreme spike deformation or prolonged high-frequency bursts of impulses) all units have responded to visual stimulation, though it has occasionally taken several hours to find the retinal region containing the receptive field and to work out the optimum stimuli. Some cells responded only to stimuli which were optimum in their retinal position and in their form, orientation and rate of
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

123
CAT VISUAL CORTEX movement. A few even required stimulation of both eyes before a response could be elicited (see Part II). But there is no indication from our studies that the striate cortex contains nerve cells that are unresponsive to visual stimuli. Most of the cells of this series were observed for 1 or 2 hr, and some were studied for up to 9 hr. Over these periods of time there were no qualitative changes in the characteristics of receptive fields: their complexity, arrangements of excitatory and inhibitory areas, axis orientation and position all remained the same, as did the ocular dominance. With deepening anaesthesia a cell became less responsive, so that stimuli that had formerly been weak tended to become even weaker or ineffective, while those that had evoked brisk responses now evoked only weak ones. The last thing to disappear with very deep anaesthesia was usually the response to a moving form. As long as any responses remained the cell retained the same specific requirements as to stimulus form, orientation and rate of movement, suggesting that however the drug exerted its effects, it did not to any important extent functionally disrupt the specific visual connexions. A comparison of visual responses in the anaesthetized animal with those in the unanaesthetized, unrestrained preparation (Hubel, 1959) shows that the main differences lie in the frequency and firing patterns of the maintained activity and in the vigour of responses, rather than in the basic receptive-field organization. It should be emphasized, however, that even in light anaesthesia or in the attentive state diffuse light remains relatively ineffective; thus the balance between excitatory and inhibitory influences is apparently maintained in the waking state.

PART II
BINOCULAR INTERACTION AND OCULAR DOMINANCE
Recording from single cells at various levels in the visual system offers a direct means of determining the site of convergence of impulses from the two eyes. In the lateral geniculate body, the first point at which convergence is at all likely, binocularly influenced cells have been observed, but it would seem that these constitute at most a small minority of the total population of geniculate cells (Erulkar & Fillenz, 1958, 1960;
Bishop, Burke & Davis, 1959; Griisser & Sauer, 1960; Hubel & Wiesel,
1961). Silver-degeneration studies show that in each layer of the geniculate the terminals of fibres from a single eye are grouped together, with only minor overlap in the interlaminar regions (Silva, 1956; Hayhow, 1958).
The anatomical and physiological findings are thus in good agreement.
It has long been recognized that the greater part of the cat's primary visual cortex receives projections from the two eyes. The anatomical
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

124
D. H. HUBEL AND T. N. WIESEL evidence rests largely on the observation that cells in all three lateral geniculate layers degenerate following a localized lesion in the striate area
(Minkowski, 1913). Physiological confirmation was obtained by Talbot &
Marshall (1941) who stimulated the visual fields of the separate eyes with small spots of light, and mapped the evoked cortical slow waves. Still unsettled, however, was the question of whether individual cortical cells receive projections from both eyes, or whether the cortex contains a mixture of cells, some activated by one eye, some by the other. We have recently shown that many cells in the visual cortex can be influenced by both eyes (Hubel & Wiesel, 1959). The present section contains further observations on binocular interaction. We have been particularly interested in learning whether the eyes work in synergy or in opposition, how the relative influence of the two eyes varies from cell to cell, and whether, on the average, one eye exerts more influence than the other on the cells of a given hemisphere.
RESULTS

In agreement with previous findings (Hubel & Wiesel, 1959) the receptive fields of all binocularly influenced cortical cells occupied corresponding positions on the two retinas, and were strikingly similar in their organization. For simple fields the spatial arrangements of excitatory and inhibitory regions were the same; for complex fields the stimuli that excited or inhibited the cell through one eye had similar effects through the other.
Axis orientations of the two receptive fields were the same. Indeed, the only differences ever seen between the two fields were related to eye dominance: identical stimuli to the two eyes did not necessarily evoke equally strong responses from a given cell. For some cells the responses were equal or almost so; for others one eye tended to dominate. Whenever the two retinas were stimulated in identical fashion in corresponding regions, their effects summed, i.e. they worked in synergy. On the other hand, if antagonistic regions in the two eyes were stimulated so that one eye had an excitatory effect and the other an inhibitory one, then the responses tended to cancel (Hubel & Wiesel, 1959, Fig. 10A).
Some units did not respond to stimulation of either eye alone but could be activated only by simultaneous stimulation of the two eyes. Text-figure
10 shows an example of this, and also illustrates ordinary binocular synergy.
Two simultaneously recorded cells both responded best to transverse movement of a rectangle oriented in a 1 o'clock-7 o'clock direction (Text-fig.
IOA, B). For one of the cells movement down and to the right was more effective than movement up and to the left. Responses fromn the individual eyes were roughly equal. On simultaneous stimulation of the two eyes both units responded much more vigorously. Now a third cell was also activated.
Downloaded from J Physiol (jp.physoc.org) by guest on August 18, 2012

CAT VISUAL CORTEX
125
The threshold of this,third unit was apparently so high that, at least under these experimental conditions, stimulation of either eye alone failed to evoke any response.
A second example of synergy is seen in Text-fig. 11. The most effective stimulus was a vertically oriented rectangle moved across the receptive field from left to right. Here the use of both eyes not only enhanced the response already observed with a single eye, but brought into the open a tendency that was formerly unsuspected. Each eye mediated a weak
A

INNs

I

Ilk-illt
*X
\~'H
7 -,N
- I,

B

8

>