1940 PROCEEDINGS OF THE IRE Novmber

What the Frog's Eye Tells the Frog's Brain *

J. Y. LETTVINt, H. R. MATURANAT, W. S. McCULLOCH||, SENIOR MEMBER, IRE,
AND W. H. PITTS|

Summary-In this paper, we analyze the activity of single fibers it moves like one. He can be fooled easily not only by a
in the optic nerve of a frog. Our method is to find what sort of stimu- bit of dangled meat but by any moving small object.
lus causes the largest activity in one nerve fiber and then what is the His sex life is conducted by sound and touch. His choice
exciting aspect of that stimulus such that variations in everything else
cause little change in the response. It has been known for the past of paths in escaping enemies does not seem to be gov-
20 years that each fiber is connected not to a few rods and cones in erned by anything more devious than leaping to where
the retina but to very many over a fair area. Our results show that for it is darker. Since he is equally at home in water and on
the most part within that area, it is not the light intensity itself but lanid, why should it matter where he lights after jumping
rather the pattern of local variation of intensity that is the exciting or what particular direction he takes? He does remember
factor. There are four types of fibers, each type concerned with a dif-
ferent sort of pattern. Each type is uniformly distributed over the a moving thing providing it stays within his field of
whole retina of the frog. Thus, there are four distinct parallel dis- vision and he is not distracted.
tributed channels whereby the frog's eye informs his brain about the
A natomy of Frog Visual A pparatus
visual image in terms of local pattern independent of average
illumination. We describe the patterns and show the functional and
The retina of a frog is shown in Fig. l(a). Between
anatomical separation of the channels. This work has been done on
the frog, and our interpretation applies only to the frog. the rods and cones of the retina and the ganglion cells,
whose axons form the optic nerve, lies a layer of con-
INTRODUCTION necting neurons (bipolars, horizontals, and amacrines).
Behavior of a Frog In the frog there are about 1 million receptors, 21 to 31
million conniecting neurons, and half a million ganglion
t FROG hunts on land by vision. He escapes cells [1]. The connections are such that there is a syn-
enemies mainly by seeing them. His eyes do not aptic path from a rod or cone to a great many ganglion
move, as do ours, to follow prey, attend suspi- cells, and a ganglion cell receives paths from a great
cious events, or search for things of interest. If his body many thousand receptors. Clearly, such an arrangement
changes its position with respect to gravity or the whole would not allow for good resolution were the retina
visual world is rotated about him, then he shows coin- meant to map an image in terms of light intensity point
pensatory eye movements. These movements enter his by point inito a distribution of excitement in the optic
hunting and evading habits only, e.g., as he sits on a nerve.
rocking lily pad. Thus his eyes are actively stabilized. There is only one layer of ganglion cells in the frog.
He has no fovea, or region of greatest acuity in vision, These cells are half a million in number (as against one
upon which he must ceniter a part of the image. He also million rods and cones). The neurons are packed to-
has only a single visual system, retina to colliculus, not gether tightly in a sheet at the level of the cell bodies.
a double one such as ours where the retina send& Thbers Their denidrites, which may extend laterally from 50,u to
not only to colliculus but to the lateral geniculatt; body 500 ,u, interlace widely into what is called the inner plexi-
which relays to cerebral cortex. Thus, we chose to work forml layer, which is a close-packed neuropil containing
oni the frog because of the uniformity of his retina, the the terminal arbors of those neurons that lie between re-
normal lack of eye and head movements except for ceptors and ganglion cells. Thus, the amount of overlap
those which stabilize the retinal image, and the relative of adjacent ganglioni cells is enormous in respect to
simplicity of the connection of his eye to his brain1. what they see. Morphologically, there are several types
The frog does not seem to see or, at any rate, is not of these cells that are as distinct ini their dendritic pat-
conicerned with the detail of stationary parts of the terns as different species of trees, from which we infer
world around him. He will starve to death surrounided that they work in different ways. The anatomy shown
by food if it is not moving. His choice of food is deter- in the figures is that found in standard referenices. Fur-
mined only by size and movement. He will leap to cap- ther discussion of anatomiiical questionis and additioinal
ture any object the size of an insect or worm, providing original work on themn will appear in a later publicationi.

* Original manuscript received by the IRE, September 3, 1959. Physiology as Known up to This Stutdy
This work was supported in part by the U. S. Army (Signal Corps),
the U. S. Air Force (Office of Sci. Res., Air Res. and Dev. Command),
Hartline [21 first used the term receptive field for the
and the U. S. Navy (Office of Naval Res.); and in part by Bell Tele- region of retina within which a local change of bright-
phone Labs., Inc. ness would cause the ganglion cell he xwas observing to
t Res. Lab. of Electronics and Dept. of Biology, Mass. Inst. Tech.,
Cambridge, Mass. discharge. Such a region is sometimes surrounded by ani
$ Res. Lab. of Electronics, Mass. Inst. Tech., Cambridge, Mass.,
on leave from the University of Chile, Santiago, Chile.
annulus, within which changes of brightness affect the
1 Res. Lab. of Electronics, Mass. Inst. Tech., Cambridge, Mass. cell's response to what is occurring in the receptive field,
1959 Lettvin, et al.: What the Frog's Eye Tells the Frog's Brain 1941

,)

(a) (b)

Tr. Opt. mod.

S
* II i
l Ur. sIb. ma. Sty. at. ent
_ _ , ,

(c) (d)
Fig. 1-(a) This is a diagram of the frog retina done by Ramon y Cajal over 50 years ago [9]. The rods and cones are the group of elements in
the upper left quarter of the picture. To their bushy bottom ends are connected the bipolar cells of the intermediate layer, for example, f, g,
and h. Lateral connecting neurons, called horizontal and amacrine cells, also occur in this layer, for example, i, j and m. The bipolars send
their axons down to arborize in the inner plexiform layer, roughly the region bounded by cell m above and the bodies of the ganglion cells,
o, p and q, below. In this sketch, Ramon has the axons of the bipolar cells emitting bushes at all levels in the plexiform layer; in fact, malny
of them branch at only one or two levels.
Compare the dendrites of the different ganglion cells. Not only do they spread out at different levels in the plexiform layer, but the pat-
terns of branching are different. Other ganglion cells, not shown here, have multiple arbors spreading out like a plane tree at two or three
levels. If the terminals of the bipolar cells are systematically arranged in depth, making a laminar operational map of the rods and cones
in terms of very local contrast, color, ON, OFF, etc., then the different shapes of the ganglion cells would correspond to different combina-
tions of the local operations done by the bipolars. Thus would arise the more complex operations of the ganglion cells as described in the
text. (b) This is Ramon y Cajal's diagram of the total decussation or crossing of the optic nerve fibers in the frog [9]. He made this picture
to explain the value of the crossing as preserving continuity in the map of the visual world. 0 is the optic nerve and C is the superior collicu-
lus or optic tectum (the names are synonymous). (c) This is Ariens-Kapper's picture of the cross section of the brain of a frog through the
collicUIus, which is the upper or dorsal part above the enclosed space. (d) This is Pedro Ramon Cajal's diagram of the nervous organization
of the tectum of a frog. The terminal bushes of the optic nerve fibers are labelled a, b, and c. A, B, C, D and E are tectal cells receiving from
the optic nerve fibers. Note that the axons of these cells come off the dendrites in stratum 7, which we call the palisade layer. The endings
discussed in this paper lie between the surface and that stratum.
1942 PROCEEDINGS OF THE IRE NVovember

although the cell does not discharge to any event oc- shows (with a delay) where brightenini,g hias occurred.
curring in the annulus alonie. Like Kuffler [4], we con- To an unchanging visual pattern, the n-erve ought to
sider the receptive field and its interacting annulus as a become fairly silent after a while.
single entity, with apologies to Dr. Hartline for the Consider the retinal image as it appears in each ol the
slight change in meaning. Hartline found three sorts of three distributed channels. For both the OFF anid ON
receptive field in the frog: ON, ON-OFF, and OFF. If chainnels, we can treat the operation on the image by
a small spot of light suddenly appears in the receptive supposing that every point on the retina gives rise to a
field of an ON-cell, the discharge soon begins, inicreases blur about the size of a receptive field. Theni the OFF
in rate to some limit determined by the intensity and chaninel tells, with a long decay time, where the blurred
area of the spot, and thereafter slowly declinies. Turniing image is darkened, and the ON channel tells with a de-
off the spot abolishes the discharge. lay and long decay where it is brightened. The third
lf the small spot of light suddenly appears or disap- channel, ON-OFF, prinicipally registers imioving edges.
pears within the field of an ON-OFF cell, the discharge Having the mental picture of anl image as it appears
is short anid occurs in both cases. through the three kinds of channel, we are still faced
If the spot of light disappears from the field of an with the question of how the aniimal abstracts what is
OFF cell, the discharge begins immediately, decreases useful to him from his surroundings. At this poinlt, a
slowly in frequency, and lasts a long time. It can be safe position would be that a fair amounit of data re-
abolished promptly by turning the spot of light on again. duction has in fact beeni accomplished by the retinla anid
For all three sorts of field, sensitivity is greatest at that the interpretation is the work of the braini, a yet-
the center of each field and least at the periphery. to-be unravelled mystery. Yet the niagging worries re-
Barlow [3] extended Hartline's observations. He ob- main: why are there two complemlenitary projectionis of
served that the OFF cells have an adding receptive equally poor resolution? Why is the mosaic of receptors
field, i.e., the response occurs always to OFF at both so uselessly fine?
center and periphery of that field, and that the effect
of removinig light from the periphery adds to the effect Initial A rgument
of a reduction of light at the center, with a weight de- The assumption has always been that the eye mainly
creasing with distance. senses light, whose local distribution is transmitted to
The ON-OFF cells, however, have differencing re- the brain in a kind of copy by a mosaic of impulses.
ceptive fields. A discharge of several spikes to the ap- Suppose we held otherwise, that the nervous apparatus
pearance of light in the center is much diminished if a in the eye is itself devoted to detecting certain patterns
light is turned on in the extreme periphery. The same of light and their changes, corresponiding to particular
interaction occurs when these lights are removed. Thus, relations in the visible world. If this should be the case,
ani ON-OFF cell seems to be measuring inequality of il- the laws found by using small spots of light on the retina
lumination within its receptive field. (Kuffler [4] at the may be true and yet, in a sense, be misleadinig. Consider,
same time showed a similar mutual antagonism be- for example, a bright spot appearing in a receptive
tween ceniter and periphery in each receptive field of field. Its actual and sensible properties include niot
ganglion cells in the eye of a cat, and later Barlow, only inten-sity, but the shape of its edge, its size, curva-
Kuffler and Fitzhugh [5] showed that the size of the ture, contrast, etc.
cat's receptive fields varied with general illumination.) We decided then how we ought to work. First, we
Barlow saw that ON-OFF cells were profoundly sensi- should find a way of recording from single myeliniated
tive to movement withini the receptive field. The ON an-d untmyelinated fibers in the intact optic nerve. Sec-
cells have niot been characterized by similar methods. ond, we should present the frog with as wide a ranige of
These findings of Hartlinie and Barlow establish that visible stimuli as we could, not only spots of light but
optic nierve fibers (the axons of the ganglion cells) do things he would be disposed to eat, other thinigs fromii
nlot tranismit information only about light intensity at which he would flee, sundry geometrical figures, sta-
single poinlts in the retina. Rather, each fiber measures tionary and moving about, etc. From the variety of
a certain feature of the whole distribution of light in an stimuli we should then try to discover what commonilioi
area of the receptive field. There are three sorts of func- features were abstracted by whatever groups of fibers
tion, or areal operation, according to these authors, so we could find in the optic nierve. Third, we should seek
that the optic nierve looks at the image on the retina the anatomical basis for the grouping.'
through three distributed channels. In any one channel,
the overlap of individual receptive fields is very great. This program had started once before in our laboratory with A.
Thus one is led to the notion that what comes to the Andrew [61, 171 of Glasgow who unfortunately had to return to Scot-
land before the work got well under way. However, he had reported
brain of a frog is this: for any visual event, the OFF in 1957 that he found elements in the colliculus of the frog that were
channel tells how much dimming of light has occurred sensitive to movement of a spot of light (a dot on an oscilloscope
screen) even when the intensity of the spot was so low that turning it
and where; the ON-OFF channel tells where the bound- on and off produced no response. In particular, the elements he ob-
aries of lighted areas are moving, or where local in- served showed firing upon movement away from the centers of their
receptive fields, but not to centripetal movements. As will appear
equalities of illuminationi are forming; the ON channel later, this sort of response is a natural property of OFF fibers.
195.9 Lettvin, et al.: What the Frog's Eye Tells the Frog's Brain 1943
(ACTUAL) METHODS a rectangle 300 long and 12° wide. However, in the
Using a variant of Dowben and Rose's platinum textual report, we use a variety of other objects. As an
indicator for the stimulus, we first used a phototube
black-tipped electrode described in another paper of looking at an image of the hemisphere taken through a
this issue, we then began a systematic study of fibers camera lens and focussed on the plane of a diaphragm.
in the optic nerve. One of the authors (H. R. M.) had (Later we used a photomultiplier, so connected as to
completed the electron microscopy of optic nerve in give us a logarithmic response over about 4 decades.)
frogs [8], and with his findings we were able to under- Thus we could vary how much of the hemisphere was
stand quickly why certain kinds of record occurred. seen by the stimulus detector and match that area in
He had found that the optic nerve of a frog contains position and size against the receptive field of the fiber
about half a million fibers (ten times the earlier esti- we were studying. The output of this arrangemeint is
mates by light microscopy). There are 30 times as many the stimulus line in the figures.
unmyelinated axons as myelinated, and both kinids
are uniformly distributed throughout the nerve. The FINDINGS
axons lie in small densely packed bundles of five to There are four separate operations on the image in
100 fibers with about 100 A between axons, each bundle the frog's eye. Each has its result tranismitted by a par-
surrounded by one or more glial cells [8]. But along the ticular group of fibers, uniformly distributed across the
nerve no bundle maintains its identity long, for the retina, and they are all nearly independent of the gen-
component fibers so braid between bundles that no two eral illumination. The operations are: 1) sustained con-
fibers stay adjacent. Thus the place a fiber crosses one trast detection; 2) net convexity detection; 3) moving edge
section of the nerve bears little relation to its origin in detection; and 4) net dimming detection. The first two are
the retina anid little relation to where it crosses another reported by unmyelinated fibers, the last two by my-
section some distance away. eliinated fibers. Since we are now dealing with events
Fibers are so densely packed that one might suppose rather than poinlt excitations as stimuli, receptive fields
such braiding necessary to prevent serious interactions. canl only be defined approximately, and one cannot nec-
On the other hand, the density makes the recording essarily distinguish concentric subdivisions. The fibers
easier. A glial wall surrounds groups rather than single reportinlg the different operations differ systematically
fibers, and penetration of the wall brings the tip among not only in fiber diameter (or conduction velocity) but
really bare axons each surrounded by neighbors whose also in rough size of receptive field, which ranges from
effect is to increase the externial impedance to its action about 2° diameter for the first operationi, to about 150
currents, augmenting the external potential in propor- for the last. The following description of these groups is
tion. Thus, while we prefer to use platinum black tips definite.
to improve the ratio of signal to noise, we recorded
much the same population with ordinary sharp micro- 1) Sustained Contrast Detectors
electrodes of bright Pt or Ag. The method records An unmyelinated axon of this group does not respond
equally well from unmyelinated and myelinated fibers. when the general illumination is turned on or off. If the
We used Rana pipiens in these experiments. We opened sharp edge of an object either lighter or darker than the
a small flap of bone either just behind the eye to expose
the optic nerve, or over the brain to expose the superior background moves into its field and stops, it discharges
promptly and continues discharging, no matter what
colliculus. No further surgery was done except to open the shape of the edge or whether the object is smaller or
the membranes of connective tissue overlying the nerv- larger than the receptive field. The sustained discharge
ous structure. The frog was held in extension to a cork can be interrupted (or greatly reduced) in these axons
platform and covered with moist cloth. An animal in by switching all light off. When the light is then restored,
such a position, having most of his body surface in the sustained discharge begins again after a pause.
physical contact with something, goes into a still re- Indeed the response to turning on a distribution of light
action-i.e., he will not even attempt to move save to furnished with sharp contrast within the field is exactly
pain, and except for the quick small incision of the skin that reported by Hartline for his ON fibers. In some
at the start of the operation our procedure seems to be fibers of this group, a contrast previously within the
painless to him. With the animal mounted, we confront field is "remembered" after the light is turned off, for
his eye with an aluminum hemisphere, 20 mils thick they will keep up a low mutter of activity that is not
and 14 inches in diameter, silvered to a matte grey present if no contrast was there before. That this is not
finish on the inner surface and held concentric to the an extraordinary sensitivity of such an element in al-
eye. On the inner surface of this hemisphere, various most complete darkness can be shown by importing a
objects attached to small magnets can be moved about contrast into its receptive field after darkening in the
by a large magnet moved by hand on the outer surface. absence of contrast. No mutter occurs then. This mem-
On our hemisphere, 10 is slightly less than an eighth of ory lasts for at least a minute of darkness in some units.
an inch long. In the tests illustrated, we use as stimu- In Fig. 2 we see the response of such a fiber in the op-
lating objects a dull black disk almost 1° in diameter and tic nerve. We compare these responses with full illumi-
 1944 PROCEEDINGS OF THE IRE November

(a) (b)

(I
11 .46v-

(C)

;.li--
i--
I_ii -A I L
-m mmi m m

(d) (e)

(f) (g) (h)

. _~~~~~~~~~~AA

(i) (i)
Fig. 2-Operation 1)-contrast detectors. The records were all taken directly with a Polaroid camera. The spikes are clipped at the lower enid
just above the noise and brightened on the screen. Occasional spikes have been intensified by hand for purposes of reproduction. The resolu-
tion is not good but we think that the responses are not ambiguous. Our alternate recording method is by means of a device which displays
the logarithm of pulse interval of signals through a pulse height pick-off. However, such records would take too much explanation anld
would not add much to the substance of the present paper. (a) This record is from a single fiber in the optic nerve. The base line is the ouit-
put of a photocell watching a somewhat larger area than the receptive field of the fiber. Darkening is given by downward deflection. l he
response is seen with the nioise clipped off. The fiber discharge to movement of the edge of a 30 black disk passed in one direction but not to
the reverse movement. (Time marks, 20 per second.) (b) The same fiber shown here giving a continued response when the edge stops in the
field. The response disappears if the illumination is turned off and reappears when it is turned on. Below is shown again the asymmetry of
the response to a faster movement. (Time marks, 20 per second.) (c) The same fiber is stimulated here to show asymmetrical responise to
the 30 black object moved in one direction, then the reverse and the stimuli are repeated under a little less than a 3-decade change of il-
lumination in two steps. The bottom record is in extremely dim light, the top in very bright light. (Time marks, 20 per second.) (d) In the
bottom line, a group of endings from such fibers is shown recorded from the first layer in the tectum. A black disk 10 in diameter is moved
first through the field and then into the field and stopped. In the top line, the receptive field is watched by a photomultiplier (see text) anld
darkening is given by upward deflection. (Time marks, 5 per second for all tectal records.) (e) OFF and ON of general illumination has lno
effect on these fibers. (f) A 30 black disk is moved into the field and stopped. The response continues until the lights are turned OFF blut
reappears when the lights are turned ON. These fibers are nonerasable. (g) A very large black square is moved into the field and stopped.
The response to the edge continues so long as the edge is in the field. (h) The 3° disk is again moved into the field and stopped. When it
leaves, there is a slight after-discharge. (i) A 1° object is moved into the field, stopped, the light is then turned off, then on, and the re-
sponse comes back. The light is, however, a little less than 300 X dimmer than in the next frame. Full ON and OFF are given in the rectani-
gular calibration on the right. (j) The same procedure as in Fig. 2(i) is done under very bright light. The return of response after reini-
troducing the light seems more prolonged-but this is due only to the fact that, in Fig. 2(i), the edge was not stopped in optimal positioi.
19.59 Lettvin, et al.: Wi7hat the Frog's Eye Tells the Frog's Brain 1945

nation (a 60-watt bulb and reflector mounted a foot moved as a whole produce less direct response to move-
away from the plane of the opening of the hemisphere) ment and much less sustained response on stopping than
to those with less than 1/300 as much light (we put a if the spots are coalesced to a single larger spot.
variable resistance in series with the bulb so that the A delightful exhibit uses a large color photograph of
color changed also). We are struck by the smallness of the natural habitat of a frog from a frog's eye view,
the resulting change. In very dim light where we can flowers and grass. We can move this photograph through
barely see the stimulating object ourselves, we still get the receptive field of such a fiber, waving it around at a
very much the same response. 7-inch distance: there is no response. If we perch with a
magnet a fly-sized object 10 large on the part of the pic-
2) Net Convexity Detectors ture seen by the receptive field and move only the ob-
These fibers form the other subdivision of the unmy- ject we get an excellent response. If the object is fixed to
elinated population, and require a number of conditions the picture in about the same place and the whole moved
to specify when they will respond. To our minds, this about, then there is none.
group contains the most remarkable elements in the Finally, the response does not depend on how much
optic nerve. darker the object is than its background, so long as it is
Such a fiber does not respond to change in general distinguishably so anid has a clear-cut edge. If a disk has
illumination. It does responid to a small object (30 or a very dark center and merges gradually into the grey
less) passed through the field; the response does not of the background at the boundary, the response to it is
outlast the passage. It continues responding for a lonig very much less than to a uniform grey disk only slightly
time if the object is imported and left in the field, but darker than the background. Objects lighter than the
the discharge is permanently turned off (erased) by a background produce almost no response unless they
transient general darkness lasting 1/10 second or have enough relief to cast a slight shadow at the edge.
loniger. We have not tried shorter obscurations. All the responses we have mentioned are relatively
The fiber will not respond to the straight edge of a independent of illumination, and Fig. 3 takeen as de-
dark object moving through its receptive field or brought scribed in the caption shows the reactions to a 30 object
there and stopped. If the field is about 71 in diameter, and the large rectangle under some of the conditions
then, if we move a dark square 80 on the side through described.
it with the edge in advance there is no response, but if
the corner is in advance then there is a good one. Usu- General Comments on Groups 1) and 2)
ally a fiber will respond indefinitely only to objects The two sorts of detectors mentioned seem to include
which have moved into the field and then lie wholly or all the unmyelinated fibers, with conduction velocities
almost wholly interior to the receptive field. The dis- of 20 to 50 cm. The two groups are not entirely distinict.
charge is greater the greater the convexity, or positive There are transition cases. On one hand, some conivexity
curvature, of the boundary of the dark object until the detectors respond well to very slightly curved edges,
object becomes as small as about 2 the width of the re- even so far as to show an occasional sustained response
ceptive field. At this point, we get the largest response if that edge is left in the field. They may also not be comii-
on moving across that field, and a good, sustained re- pletely erasable (though very markedly affected by atn
sponse oni entering it and stopping. As one uses smaller interruption of light) for small objects. Oni the other
atnd smaller objects, the response to moving across the hand, others of the same group will be difficult to set
field begins to diminish at a size of about 10, although into an indefinitely sustained response with any object,
the sustained response to coming in and stoppinig re- but only show a fairly long discharge, acting thereby
mains. In this way we find the smallest object to which more as detectors of edges although never reacting to
these fibers respond is less than 3 minutes of arc. A straight edges. Nevertheless the distribution of the un-
smooth motion across the receptive field has less effect myelinated axons into two groups is very marked. AIny
than a jerky one, if the jerks recur at intervals longer fiber of either group may show a directionial responise-
than 1 second. A displacement barely visible to the ex- i.e., there will be a direction of movement that may fail
perimenter produces a marked increase in response to excite the cell. For the contrast fibers, this will also
which dies down slowly. appear as a nonexciting angle of the boundary with
Any checked or dotted pattern (in the latter case, respect to the axis of the frog. Such null directionis and
with dots no further apart than half the width of the re- angles cancel out in the aggregate.
ceptive field) moved as a whole across the receptive field
produces little if any response. However, if any dot 3) Moving-Edge Detectors
withini the receptive field moves differentially with re- These fibers are myelinated and conduct at a velocity
spect to the background pattern, the response is to that in the neighborhood of 2 meters per second. They are
dot as if it were moving alone. A group of two or three the same as Hartline's and Barlow's ON-OFF units.
distiiict spots enclosed within the receptive field and The receptive field is about 12° wide. Such a fiber re-
1946 PROCEEDINGS OF THE IRE November

r,
lingii- ~(h
'k
(a) (b)

. A _

| *Ai -.Mmd. -MiIIIII.,li

- T I

(c) (d)

0
t

i
t

(e) (f)
~~~~~( f )
~~A~I4 a*.

(g) (h)
Fig. 3-Operation 2)-convexity detectors. The photomultiplier is used, and darkening is an upward deflection. (a) These records are all fromn
the second layer of endings in the tectum. In the first picture, a 10 black disk is imported into the receptive field and left there. (b) The
saiime event occurs as in Fig. 3(a), but now the light is tturned off then on again. The response is much diminished and in the longer record
v.taiishes. These fibers are erasable. (c) The 10 disk, is passed through the field first somewhat rapidly, then slowly, then rapidly. The light
is xvery bright. (d) The same procedure occurs as in Fig. 3(c), but now the light has been dimmed about 300X. The vertical line shows
the range of the photomultiplier which has been adjusted for about 3' decades of logarithmic response. (e) A 1° black disk is passed
through the field at three speeds. (f) A 150 black strip is passed through at two speeds edge leading. (g) A 15° black strip is passed throlugh
in various ways with corner leading. (h) The same strip as in Fig. 3(g) is passed throtugh, edge leading.

sponids to any distinguishable edge moving through its clocked at 10 meters per second.2 One such fiber re-
receptive field, whether black against white or the other sponids to sudden reduction of illum-iination by a pro-
way around. Chaniging the extent of the edge makes longed anid regular discharge. Indeed, the rhythm is so
little difference over a wide range, changing its speed a much the same fromi fiber to fiber that in recording from]
great one. It responds to anl edge oiily if that edge several at onice after sudden darkening, the impulses as-
moves, not otherwise. If an object wider thani about 50 semlble in groups, which break up onily after many sec-
moves smnoothly across the field, there are separate re- onds. Eveni then the activity from widely separated
sponses to the leading and trailinig edges, as would be ex- retinal areas seems to be related. We observe that the
pected from Barlow's formulation. These fibers geni- surface potential of the colliculus shows a violeint and(I
erally show two or three regularly spaced spikes, always proloiiged oscillation when the light is turned off. This
synlchronous amiiong different fibers to turning the light oscillation, beginninig at about 18 per second aiid breakl-
oni or off or both. The response to mnoving objects is ing into 3 to 5 per secon(l aIfter several seconds, seemiis to
imtich greater than to changes in total illumination and arise from these fibers fromii the retina; the samne reco-d
varies only slightly with general illumination over a is seen when the optic nerve is severed and the record-
range of 1/300. The frequency of the discharge increases ing electrode placed on the retinial stumiip. See Fig. 5.
with the velocity of the object withini certaini limits (see The receptive field is rather large about 15 -and
Fig. 4). works as Barlow describes. Darkeniing of a spot pro-
duces less responise wheni it is in the periphery of hlle
4) Net Dimming Detectors field than when it is at the center. The effect of a milov-
These are Hartline's and Barlow's OFF fibers. But
The even faster fibers, with velocities up to 20 mieters per secoiid,
tlhey have some properties not observed before. They we 2presenitly believe to be the efferents to the retina, but althouigh
are imiyeliiiated anid the fastest conducting afferents, there is some evidence for this, we are not yet qtuite certaiui.
1959 Lettvin, et al.: What the Frog's Eye Tells the Frog's Brain 1947

(a

(a) (b)

--I I I __I
I L

6,-

(c) (d)

_ .bL

()
(e) (f)

(g) (h)
Fig. 4-Operation 3)-moving-edge detectors. The first two pictures are taken from a single fiber in the optic nerve. (a) Shows a 7° black
disk moving through the receptive field (the photocell was not in registration with the field). There is a response to the front and back
of the disk independent of illumination. There is about a 300,/1 shift of illumination between top and bottom of the record. Darkening is a
downward deflection with the photocell record. (Time marks, 5 per second.) (b) OFF and ON of general lighting. (Timie marks, 50 per
second.) Note double responses and spacing. (c) This and succeeding records are in the third layer of endings in the tectlum. Several endings
are recorded but not resolved. Darkening is an upward deflection of the photomultiplier record. The responise is shown to the edge of a 150
square moved into and out of the field by jerks in bright light. (d) The same procedure occurs as in Fig. 4(c), but in dim light. Calibra-
tion figure is at the right. (e) The response is shown to a 7° black disk passed through the receptive fields under bright light. The sweep is
faster, btut the time marks are the same. (f) The same procedure as for Fig. 4(e), but under dim light. The OFF and ON of the photo-
multiplier record was superimposed for calibration. (g) OFF and ON response with about half a second between ON and OFF. (h) Same as
Fig. 4(g), but with 2 seconds between OFF and ON.

ing object is directly related to its size and relative dark- terrupt the response gets less and less the longer onie
ness. The response is prolonged if a large dark object waits. That is, the sensitivity of the OFF discharge to
stops within the field. It is almost independent of il- the ON of light increases with time. If we darkein the
lumination, actually increasing as the light gets dim- general lighting onlv by a factor of 100, we also get a
mer. There is a kind of erasure that is complementary proloniged discharge. However, if we turn off the light
to that of group 2). If the general lighting is sharply completely a few seconds after the 100/1 dimming and
dimmed, but not turned off entirely, the consequent then turn it back on to the same dim level, the dis-
prolonged response is diminished or abolished after a charge is increased by the second dimmintg and is com-
dark object passes through the receptive field. In this pletely or almost completely abolished by the relight-
case, the reasotns for erasure are apparent. Suppose one ing. The effect of moving a dark object through the field
turns off the light and sets up a prolonged response. after dimming is to impose a second dimming pulse fol-
Then the amounit of light which must be restored to in- lowed by brightening as the object passes.
1948 PROCEEDINGS OF THE IRE November

(a) (b)

- r
,
'I-
q- - - T

(c) (d)
I,; 1- 11.11

I
k
i I
priiii
-.L ,
. Ili I, bIL,
--
-lI-i-- --
I
I W

(e) (f)
"LiLl. I it, diall I i I ui .,LI I" I, II Li ..L- I. A JL L.
--
77--, 7--T, -'I T- --y- .

(g)
Fig. 5-Operation 4)-dimming detectors. (a) This and the next framne are taken from a single fiber in the optic ilerve. Here we see the
response to a 7° black disk passing through the receptive field. The three records are taken at three illuminatioll levels over a 300:1 range.
In the phototube record, darkening is a downward deflection. (Time marks, 5 per second.) (b) OFF and ON of light. The OFF was dolne
shortly after one sweep began, the ON occurred a little earlier on the next sweep. The fiber is silenced completely by the ON. (Time
marks, 5 per second.) (c) In this and the next three frames, we are recording from the fourth layer of endings in the tectum. This frame
shows the response to turning OFF the general illumination. (d) OFF and ON of light at regular intervals. (e) OFF then ON of the light
to a lesser brightness. (f) OFF then ON of the light to a still lesser brightness. The level to which the ON must come to abolish activity
decreases steadily with time. (g) The synchrony of the dimming detectors as described in the text. At the top are three or four fibers
recorded together in the optic nerve when the light is suddenly turned off. The fibers come from diverse areas on the retina. In the second
line are the oscillations recorded from the freshly cut retinal stunip of the optic nerve when the light is. suddenly turned off. In the third
line are the oscillations recorded on the surface of the tectum, the visual brain, before the nerve was cut. Again the light is suddenly
turned off. The last line is 20 cps. These records of synchrony were obviously not all made at the same time, so that comparing them in
detail is niot profitable.

Others Suppose that an- object is miioved about in this coni-

Lastly, there is a small group of afferent fibers which centric array:
does not seem to have distinct receptive fields. They 1) The contrast detector tells, in the smiiallest area of
each measure the absolute degree of darkness over a all, of the presence of a sharp boundary, Imlovinlg or still,
wide area with a long time constant. That is, the fre- with much or little contrast.
quency of discharge is greater the darker it is. They 2) The convexity detector iniformis us in a somnewhat
have a complement in that somiie of the moving edge de- larger area whether or not the object lhas a curved
tectors have a resting discharge of very low frequenlcy if bounidary, if it is darker than the backgrounid anid mlov-
the illumination is extremely bright. inig oIn it; it remlemiibers the object whein it has stopped,
providing the boundary lies totally withini that area and
DISCUSSION is sharp; it shows most activity if the eniclosed object
Let compress
us all of these findings the
in following moves intermittently with respect to a backgroun-d. The
description. Conisider that we have four fibers, one from imiemiiory of the object is abolished if a shadoxv obscLnres
each group, which are conicentric in their receptive fields. the object for a moment.
1959 Lettvin, et al.: What the Frog's Eye Tells the Frog's Brain 1949
3) The moving-edge detector tells whether or not diameter. If the active elements in the upper layers are
there is a moving boundary in a yet larger area within cells (one will see about 20 to 30 active elemenits in one
the field. electrode track before reaching the cell layer), which
4) The dimming detector tells us how much dimming cells can they be? Or if they are dendrites, to what
occurs in the largest area, weighted by distance from neurons do they belong? We regard these considera-
the center and by how fast it happens. tions as conclusive enough.
All of the operations are independent of general illu- Figs. 2-5 show that the four operational groups of
mination. There are 30 times as many of the first two de- fibers terminate in four separate layers of terminals,
tectors as of the last two, and the sensitivity to sharp- each layer exhibiting a continuous map of the retina
ness of edge or increments of movement in the first two (we confirm Gaze's diagram of the projection [10]) and
is also higher than in the last two. all four maps are in registration. Most superficial lie the
endings for the contrast detectors, the slowest fibers.
RESULTS IN THE TECTUM Beneath them, but not so distinctly separate, are the
As remarked earlier, the optic nerve fibers are all dis- convexity detectors. Deeper, and rather well separated,
ordered in position through the nerve. That is, the prob- are the moving-edge detectors admixed with the rare
ability that any two adjacent fibers look at the same and ill-defined axons that measure the actual level of
region in the retina is very small. However, when the darkness. Deepest (and occasionally contaminated with
fibers terminate in the superior colliculus they do so in tectal cells or their axons) lie the dimming detectors.
an orderly way such that the termini exhibit a continu- Thus the depth at which these fibers end is directly re-
ous map of the retina. Each optic nerve crosses the base lated to their speed of conduction.
of the skull and enters the opposite tectum [Fig. 1(b)] Such an arrangement makes experiment easy, for all
via two bundles-one rostromedial, the other caudalat- the fibers of one operation performed on the same field
eral. The fibers sweep out over the tectum in the super- in the retina end in one place in the tectum and can be
ficial neuropil in what grossly appears to be a laminated recorded as a group. It is very useful to see them this
way [Fig. 1 (c) ]. The detail of ending is not known, and way, for then the individual variations among similar
there is some reason to think Pedro Ramon's drawing unlits cancel one another and only the common proper-
[9] is too diagrammatic [Fig. 1(d) ], hcwever well it fits ties remain. We made the tectal records shown in the
with our data. accompanying figures with a single electrode in two
In any case, the outer husk of neuropil, roughly about penetrations (to get decent separation of contrast and
half the thickness of the optic tectum, is formed of convexity detectors which lie just below the pia), to
the endings of the optic fibers mixed with dendrites of show how clear-cut the arrangement is.
the deeper lying cells, and in this felting lie few cell
bodies. CONFIRMATION OF SPERRY'S PROPOSAL
We have found it singularly easy to record from these The existence of a fourfold map of the retina in the
terminal bushes of the optic fibers. That is, if an elec- tectal neuropil led us, naturally, to repeat Sperry's
trode is introduced in the middle of one bush, the exter- initial experiment on the regeneration of cut optic nerve
nal potential produced by action currents in any [11]. Since the nerve is as scrambled as it can be orig-
branch increases in proportion to the number of inally, we saw no point in turning the eye around 1800
bralnches niear the electrode. Since the bushes are but simply cut one nerve in a few frogs, separated the
densely initerdigitated everywhere, it is not difficult to stumps to be sure of complete severance, an-d waited
record from terminlal arbors anywhere unless one kills for about 3 months. At the end of this time, after it was
or blocks them locally, as is easily done by pressure, etc. clear that the cut nerves were functioning again, we
Onie may inquire how we can be sure of recording from compared the tectal maps of the cut and unlcut nierves
terminial arbors, and not from cells and their dendrites. in some of them. We confirmed (as did Gaze [12])
Thle argunment is this. First, there are about four layers Sperry's proposal that the fibers grew back to the re-
of cells in the depths of the tectum [Fig. 1(d) ], and only gions where they originally terminated in mapping the
their denidrites issue into the superficial neuropil wherein retina [13]. But we also found a restoration of the four
lie very few cells indeed. There are about 250,000 of layers with no error or mixing. Ill one frog, after 90 days,
these cells in all, compared to 500,000 optic fibers. In the fibers had grown back best at the enitrance of the two
the outer thickniess of the tectum, among the terminiat- brachia to the colliculus, and least at the ceniter, yet
ing fibers, almost every elemenit we record performns onie there were no serious errors. The total area of retinia
of the four operations characterizing the fibers in the comnmunicatiing with one point of the collicular nieuropil
nerve, and has a corresponding receptive field. Now as (i.e., the sum of the receptive fields of the fibers re-
the electrode moves down from the surface in one track, corded from that point) had increased three or four
we record 5 to 10 cells in the deepest half of the tectum. times, from a diameter of about 150 to a dianmeter of
Not a sinigle cell so recorded shows activity eveni re- about 300. But there was no admnixture of fibers with
motely resembling what we find in the superficial neuro- receptive fields in widely separated regionis. In another
pil. Amonig the cells, none show optic nerve operations, frog, after 120 days, the area seeni from one point was
and the smcallest receptive fields we find are over 30° in barely twice normal.
1950 PROCEEDINGS OF1IIE IRE NYovember

GENEI-RAL DIscUSSION here. The differeint denidritic distribution of the gan-

What are the colnsequenices of this work? Funida- glion cells, as in Fig. 1(a), may sign-ify that they extract
mnentally, it shows that the eye speaks to the brain in a (lifferenitly weighted samiples of simiiple local operations
language already highly organized and initerpreted, il- done by the bipolars, anid it is oni this that we are niow
stead of transsmittinig sonme imore or less accurate copy of worki n1g.
the distributioni of light oni the receptors. As a crude BLut there is aniother reasoni for a reluctanice to niake
analogy, suppose that we have a man watchinig the accurate mieasuremienits on the activity of ganiglioni cells
clouds and reportinig them to a weather stationi. If he is in the intact animal. A significanit efferent outflow goes
using a code, an-d onie can see his portion of the sky too, to the retinia fromii the brain. We nlow know to a certaill
then it is niot difficult to find out what he is saying. It is extenit how the cells in the tectumii hanidle the four in-
certainly true that he is watchin-g a distributioni of light; ptuts to themii which are described in this paper. There
nevertheless, local variations of light are not the termiis aIre at least two distinct classes of these cells, aniid att
in which he speaks nor the terms in which he is best un- leatst onie of themii issues axonls back inito the optic nierve.
(lerstood. Iindeed, if his vocabulary is restricted to types We have recorded this atctivity there. Such axonis enlter
of things that he sees in the sky, trying to find his lan- the retinia Cand we think somiie effects of their atctivity
guage by using flashes of light as stimuli will certainlly on1 the senisitivity of gan-gliol1 cells are noticeable.
fail. Now, sinice the purpose of a frog's vision is to get TIhe way that the retinia projects to the tectumil sug-
him food anid allow him to evade predators nlo miatter gests a ninleteenith cenitury view of visual space. The
how bright or dimii it is about himii, it is not eniough to imaiiige oni the retinia, takeni at the grain of the rods anld
know the reaction of his visual systemn to points of light. cones, is ani array of regularly spaced points at each ol
To get useful records from inidividual receptors (the rods which there is a certaini amiiounit of light of a certaini
and cones), assumiiing that they operate indepenidenltly comllpositioni. If we know the positioIn of every poinlt
and under no reflex control, this stimulus miay be ade- anid the values of light at every point, we can-l physi-
quate. But when onie inspects responises that are a few cally reconistruct the image, anid looking at it unider-
nervous transformiiations removed fromn the receptors, as stand the picture. If, however, we are re(luire(I to estab-
in the optic nerve, that same choice of stimulus is diffi- lish conitinuities withini the picture onily fromi the iiinulner-
cult to defend. It is equivalenit to assuminiiig that all of ical data on position anid light at inidepenidenit poilnts, it
the interpretation is donie further oni in the niervous becomiies a very difficult task. The retinia projects oiito
system. But, as we have seeni, this is lalse. the tectum in four parallel sheets of enidinigs, each sheet
One might attempt to mneasure numierically how the mappinig continuously the retina in termls of a palrticLi-
response of each kind of fiber varies with various prop- lar areal operation, anid all four maps are in registratiol).
erties of the successionis of patternis of light which evoke Conisider the dendrite of a tectal cell extenidinig ul)
them. But to characterize a succession of patterns in through the four sheets. It is looking at a point in the
space requires knowledge of so miianly independenit vari- image on the retinia, but that point is nlow seen in termiis
ables that this is hardly possible by experimiienital eniu- of the properties of its nieighborhood as definied by the
merationi of cases. To examine the effect of curvature operationis. Since the overlap of receptive fields withill
alone we should have to explore at least the response to any operation is very great, it now seems reasonable to
all configurations of three spots movinig in a large va- erect simple criteria for findinig conitinuities. I or ex-
riety of ways. We would prefer to state the operationis of amiiple, if an area over which there is little chanige in thle
ganglion cells as simply as possible in terms of whatever fourfold signiature of a ilmoving object is bounided by
quality they seem to detect anid, itnstead, examiinle the regions of different signature, it seemiis likely that that
bipolar cells in the retina, expecting to find there a dis- area describes the image of a single object.
section of the operationis inito combinations of simllpler By transforminig the imiiage fromii at space of simple
ones performed at intermediate stages. For example, discrete points to a congruent spalce where each equival-
suppose that there are at least two sorts of rods and lenit point is described by the initersectioni of particular
cones, one increasin-g its voltage with the log of light (lualities in its nieighborhood, we cani theni give the
at one color, the other decreasinig its voltage with the image in terms of distributionis of combinations of those
log of light at somiie other color. If bipolars conniiect to qualities. In short, every point is seeln inl definiite coni-
several conitiguous rods or conies of opposinig reactionis texts. Trhe character of these conitexts, genietically buLilt
and simply add voltages, somiie bipolars will register- a in, is the physiological synthetic a priori. The ol)era,-
large signal onily if an appropriate conitrast occurs. We tionis found in the frog miakze unilikely later processes in
have in fact founid somiiething of the sort occurring, for his systemll of the sort described by two of Us earlier [15 ],
it seems that the innier plexiform layer of the retina is for example, dilatations; but those were adduced for
stratified to display several differenit local properties, the sort of fornm recogniition which the frog does nlot
one layer indicatin-g local differeinces in initenisity of have. This work is ail outgrowth of that earlier study
light. Somiie of Svaetichini's [141 data catn be adduced xvhich set the questioni.
1 959 Lettvin, et al.: What the Frog's Eye Tells the Frog's Brain 1951
CONCLUSION orstill, is in the field. Could one better describe a sys-
The output from the retina of the frog is a set of four tem for detecting an accessible bug?
distributed operations on the visual mage. These opera-
tions are independent of the level of general illumina- ACKNOWLEDGMENT
tion and express the image in terms of: 1) local sharp We are particularly grateful to 0. G. Selfridge whose
edges and contrast; 2) the curvature of edge of a dark experiments with mechanical recognizers of pattern
object; 3) the movement of edges; and 4) the local dim- helped drive us to this work and whose criticism in part
mings produced by movement or rapid general darken- shaped its course.
ing. Each group of fibers serving one operation maps
the retina continuously in a single sheet of endings in BIBLIOGRAPHY
the frog's brain. There are four such sheets in the brain, [1] H. R. Maturana, "Number of fibers in the optic nerve and the
corresponding to the four operations, and their maps number of ganglion cells in the retina of Anurans," Nature, vol.
are in registration. When all axonal connections be- 183, pp. 1406-1407; May 16, 1959.
tween eye and brain are broken and the fibers grow [21 H. K. Hartline, "The response of single optic nerve fibres of the
vertebrate eye to illumination of the retina," Amer. J. Physiol.,
back, they reconstitute the original retinal maps and vol. 121, pp. 400-415; February, 1938.
Also, "The receptive fields of the optic nerve fibers," Anmer. J.
also arrange themselves in depth in the original order Physiol., vol. 130, pp. 690-699; October, 1940.
with no mistakes. If there is any randomness in the [31 H. B. Barlow, "Summation and inhibition in the frog's retilna,"
J. Physiol., vol. 119, pp. 69-88; January, 1953.
connections of this system it must be at a very fine level [4] S. W. Kuffler, "Discharge patterns and functional organization
indeed. In this, we consider that Sperry [11] is com- of mammalian retina," J. Neurophysiol., vol. 16, pp. 37-68;
January, 1953.
pletely right. [5] H. B. Barlow, R. Fitzhugh, and S. WV. Kuffler, "Change of or-
We have described each of the operations on the ganization in the receptive fields of the cat's retina during dark
adaptation," J. Physiol., vol. 137, pp. 338-354; August, 1957.
retinal image in terms of what common factors in a [6] A. M. Andrew, "Report on Frog Colliculus," Res. Lab. of Elec-
large variety of stimuli cause response and what com- tronics, Mass. Inst. Tech., Cambridge, Quarterly Progress Rept.,
pp. 77-78; July 15, 1955.
mon factors have no effect. What, then, does a par- [7] A. M. Andrew, "Action potentials from the frog colliculus,"
ticular fiber in the optic nerve measure? We have con- J. Physiol., vol. 130, p. 25P; September 23-24, 1955.
sidered it to be how much there is in a stimulus of that [81 H. R. Maturana, "The Fine Structure of the Optic Nerve and
Tectum of Anurans. An Electron Microscope Study," Ph.D. dis-
quality which excites the fiber maxin'ally, naming that sertation, Harvard University, Cambridge, Mass.; 1958.
quality. [91 Pedro Ramon Cajal, "Histologie du Systeme Nerveux," Ramon
y Cajal, Maloine, Paris, France; 1909-1911.
The operations thus have much more the flavor of [10] R. M. Gaze, "The representation of the retina on the optic lobe
of the frog," Quart. J. Exper. Physiol., vol. 43, pp. 209-214;
perception than of sensation if that distinction has any March, 1958.
meaning now. That is to say that the language in which [11] R. Sperry, "Mechanisms of neural maturation," in "Handbook
of Experimental Psychology," S. S. Stevens, Ed., John Wiley
they are best described is the language of complex ab- and Sons, Inc., New York, N. Y.; 1951.
stractions from the visual image. XVe have been tempted, [121 R. M. Gaze, "Regeneration of the optic nerve in Xenopus laevi,"
J. Physiol., vol. 146, p. 40P; February 20-21, 1959.
for example, to call the convexity detectors "bug per- [13] H. R. Maturana, J. Y. Lettvin, W. S. McCulloch, and W. H.
ceivers." Such a fiber [operation 2] responds best when Pitts, "Physiological evidence that cut optic nerve fibers in the
frog regenerate to their proper places in the tectum," Science;
a dark object, smaller than a receptive field, enters that 1959 (in press).
field, stops, and moves about intermittently thereafter. [14] G. Svaetichin and E. F. NcNichol Jr., "Retinal mechanisms for
chromatic and achromatic vision," Ann. N. Y. Arad. Sci., vol.
The response is not affected if the lighting changes or if 74, pp. 385-404; November, 1958.
the background (say a picture of grass and flowers) is [15] WV. S. McCulloch and W. H. Pitts, "How we know universals.
The perception of auditory anid visual forms," Bull. MIath. Bio-
moving, and is not there if only the background, moving physics, vol. 9, pp. 127-147; June, 1947.