J Neurophysiol 94: 3443–3450, 2005.
First published August 3, 2005; doi:10.1152/jn.00407.2005.
Laminar Variation in Threshold for Detection of Electrical Excitation of
Striate Cortex by Macaques
Edgar A. DeYoe, Jeffrey D. Lewine, and Robert W. Doty
Department of Neurobiology and Anatomy, University of Rochester School of Medicine and Dentistry, Rochester, New York
Submitted 20 April 2005; accepted in final form 23 July 2005
restricted the allowable density of stimulated loci. It was then
found, however, that macaques could respond to stimulation of
as little as 2 ␮A with microelectrodes tracking through striate
cortex (Bartlett and Doty 1980). After a successful exploration
with microelectrodes in three sighted patients by Bak et al.
(1990), Schmidt et al. (1996) proceeded to implant an array of
microelectrodes in a blind volunteer. “Two-point discrimination” was some fivefold better than what had previously been
achieved with surface stimulation, and thresholds for phosphene elicitation were comparable to stimulus levels detectable
by macaques. Bradley et al. (2005) and Tehovnik et al. (2005)
have continued to explore this approach, using an implanted
array or movable microelectrodes in sighted macaques, and
testing the animal’s ability to direct its gaze to the point in the
visual field corresponding to the stimulated locus.
Our approach herein has been to explore the circuitry responsible for the putative phosphene in macaques. Because the
animals find stimulation at all striate loci to be qualitatively
equivalent (Bartlett et al. 2005), as is the human case, the
profile of excitability by microelectrodes traversing the distinct
cortical laminae may begin to define the elements involved. At
the least, these observations, combined with findings in the
companion paper (Bartlett et al. 2005), yield a prescription for
generating detectable sensations with minimal currents and
provide key insights into the neural mechanisms responsible
for the putative phosphene sensation.
METHODS
INTRODUCTION
Animals and behavioral task
Electrical stimulation of human striate cortex evokes the
sensation of a unitary point of light—a phosphene—that usually is qualitatively invariant despite wide adjustments in the
parameters and locus of stimulation (Bak et al. 1990; Brindley
1971; Brindley and Lewin 1968; Dobelle and Mladejovsky
1974; Henderson et al. 1979; Pollen 1975; Rushton and Brindley 1978; Schmidt et al. 1996). This encourages the idea of
applying such stimulation prosthetically to the striate cortex of
the blind, excitation of multiple loci, thus allowing the evocation of visual patterns. Such usage would demand limitation of
the size of each phosphene and thus presumably, the size of the
excited neuronal field, to obtain a large number of nonoverlapping loci.
In most of these initial studies stimuli were applied to the
pial surface and required rather high levels of excitatory
current. This resulted in somewhat diffuse phosphenes, and
Eight Macaca nemestrina were used, under veterinary care. The
Guiding Principles in the Care and Use of Animals of the American
Physiological Society were rigorously followed in all instances.
The general procedures have been detailed in the companion paper
(Bartlett et al. 2005). In brief, the animal placed its hand on a 13 ⫻
20-cm capacitatively tuned plate, proper contact being signaled by a
low, continuous tone. Stimulation was presented at some random
interval between 2 and 30 s after contact, and if the animal broke
contact within 300 –1,300 ms thereafter, it was rewarded with a small
quantity of fruit juice. The time window for correct detection was set
according to the animal’s previously determined reaction times and,
for a 1-s stimulus, was typically between 300 and 1,300 ms. Failure to
respond to the stimulus within this time frame, or if contact was
broken at inappropriate times, activated a loud horn.
Because obtaining a “psychophysical function” (see Bartlett et al.
2005) at each point in the traverse of the microelectrode was impractical, threshold at a tested locus was defined as the lowest current
Address for reprint requests and other correspondence: R. W. Doty, Department
of Neurobiology and Anatomy, Box 603, University of Rochester Medical
Center, Rochester, NY 14642 (E-mail: [email protected]).
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked “advertisement”
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
www.jn.org
0022-3077/05 $8.00 Copyright © 2005 The American Physiological Society
3443
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
DeYoe, Edgar A., Jeffrey D. Lewine, and Robert W. Doty. Laminar variation in threshold for detection of electrical excitation of
striate cortex by macaques. J Neurophysiol 94: 3443–3450, 2005.
First published August 3, 2005; doi:10.1152/jn.00407.2005. Macaques were trained to signal their detection of electrical stimulation
applied by a movable microelectrode to perifoveal striate cortex.
Trains of ⱕ100 cathodal, 0.2-ms, constant current pulses were delivered at 50 or 100 Hz. The minimum current that could be reliably
detected was measured at successive depths along radial electrode
penetrations through the cortex. The lowest detection thresholds were
routinely encountered when the stimulation was applied to layer 3,
particularly just at the juncture between layers 3 and 4A. On the
average, there was a twofold variation in threshold along the penetrations, with the highest intracortical thresholds being in layers 4C
and 6. Variations as high as 20-fold were obtained in some individual
penetrations, whereas relatively little change was observed in others.
The minimum detectable current was 1 ␮A at a site in layer 3, i.e.,
10 –100 times lower than that for surface stimulation. Because macaques, as do human subjects, find electrical stimulation of striate
cortex to be highly similar at all loci (a phosphene in the human case),
it is puzzling as to how such uniformity of effect evolves from the
exceedingly intricate circuitry available to the effective stimuli. It is
hypothesized that the stimulus captures the most excitable elements,
which then suppress other functional moieties, producing only the
luminance of the phosphene. Lowest thresholds presumably are encountered when the electrode lies among these excitable elements that
can, with higher currents, be stimulated directly from some distance or
indirectly by the horizontal bands of myelinated axons, the stria of
Baillarger.
3444
E. A. DEYOE, J. D. LEWINE, AND R. W. DOTY
detected in three of five presentations. Although this method does not
yield an unbiased measure of the detection threshold (McNicol 1972),
it could be deduced within 25–30 presentations and was reliably
repeatable. Of course, changes in the animal’s criterion, attention, or
motivation could introduce error. To guard against such unrecognized
fluctuations, permanently implanted macroelectrodes were used
throughout each session to determine the threshold for a standard,
invariant stimulation. So long as this threshold remained constant, it
was assumed that changes in threshold from one to another microelectrode site reflected true differences in detectability, and were not
attributable to changes in the animal’s behavioral inclination.
Electrodes, aseptic precautions, and surgery
Electrical stimulation and recording
As indicated by Ranck (1975), the change in membrane potential at
a cell stimulated by an extracellular electrode is proportional to the
stimulus current rather than to the stimulus voltage. Consequently, the
typical stimulus consisted of cathodal current pulses delivered in a 1-s
burst of 100 pulses of 0.2 ms. The stimuli were generated by a
digitally controlled pulse generator (Pulsar 6B, FHC) optically coupled to a constant-current amplifier (FHC CCIU-8). This unit was
modified to obtain more precise stimulus waveforms at low amplitudes. An additional low-current range was added (1 ␮A/V) and the
operational amplifier-integrated circuits in the output stage were
replaced with circuits having a faster slew rate so that very short
stimulus pulses (10 ␮s) could be produced. Calibration of the digitally
controlled stimulus amplitude was monitored by observing the voltage
produced across a precision 1.00-k⍀ resistor in series with the current
return path. To avoid potentially harmful polarization of the stimulating electrodes, an exhauster circuit (Doty and Bartlett 1981; availJ Neurophysiol • VOL
Verification of microelectrode locations
Establishing an accurate association between the stimulation/recording sites and the cortical lamination was difficult, especially given
the variable compression of the superficial cortical layers as the
electrode was advanced through the intact dura mater. As one aid
to this alignment process, electrolytic lesions, 10 –15 ␮A DC for
10 –15 s, were placed at the terminus of an electrode’s advancement
or at identified positions along the track during retraction. Such
currents produce lesions of a few hundred microns in diameter, which
proved to be a satisfactory compromise between accurate localization
versus preservation of the lesion site for ⱕ6 wk.
To recover these lesions and reconstruct the electrode tracks,
euthanasia was induced with a lethal dose of barbiturate. After cardiac
perfusion with 4% formalin, the brain was removed and blocked in the
plane of the microelectrode tracks. Frozen brain sections were cut at
a thickness of 60 ␮m, mounted, and then stained with cresyl violet.
Finished slides were placed in a photographic enlarger and projected
onto paper for sketching cortical laminar boundaries and lesion sites.
In addition, quantitative measurements of track length and lesion
depths were made directly from the slides using a microscope
equipped with a calibrated eyepiece reticle.
Marking lesions were only partially useful in establishing the
laminar registration. They invariably elevated detection thresholds
within ⱖ500 ␮m of a lesion site and, consequently, could not be used
to mark multiple points during the electrode’s advancement. To
provide additional data for aligning tracks and laminae, background
multiunit activity and flash-evoked potentials were recorded, both of
which show unique changes when advancing from layer 4 into layer
5. At this border, the amplitude of the background activity drops
substantially, and the evoked potential shows an inversion of the early
N1 and P3 waves (Snyder et al. 1979). Thus the depth at which these
changes occur provides a laminar reference point within the cortex.
This is especially useful when, as in the present experiments, it is not
feasible to use single-unit receptive field properties to characterize
different laminae.
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
Macroelectrodes, fashioned from 127-␮m Teflon-coated platinum
alloy wire (92% Pt, 8% W), cuffed, and cut to a 45 ° chisel point (Doty
and Bartlett 1981), were implanted in striate cortex using full aseptic
precautions while the animal was at a surgical level of anesthesia,
maintained with halothane or intravenous pentobarbital. At the same
time, to provide comfortable stabilization of the animal’s head during
recording sessions with microelectrodes, a half-circle of aluminum
was secured to keyholes in the skull, as described by Evarts (1968).
Subsequently, during an experiment and after the animal had assumed
a posture to its liking, the aluminum piece could be clamped, thus
painlessly and temporarily preventing the animal from making sudden
movements of its head.
Using surgical-grade methacrylate, a 1-cm diameter Nylon recording chamber (FHC, Bowdoinham, ME) was also cemented to the skull
surrounding a trephine hole over opercular striate cortex, leaving the
dura mater intact. This chamber was filled through a pressure relief
port with sterile mineral oil or petroleum jelly. When used in an
experimental session, the cover of the chamber was removed after first
cleansing its surface with 70% isopropyl alcohol (being meticulously
careful, of course, not to permit contact of the alcohol with skin
around the implant) followed by H2O2. Strict asepsis was maintained
for the interior of the chamber and no infection occurred.
The microelectrodes were held in a miniature hydraulic microdrive
(FHC) accommodated to the Nylon chamber, which was refilled with
sterile mineral oil to provide a gas-free seal over the dura mater. The
platinum–iridium, glass-insulated microelectrodes (FHC) had exposed, conically etched tips roughly 10 ␮m long and 5 ␮m at the base
and, after penetrating the dura mater, had an impedance of 0.5–1.5
M⍀. They readily recorded single or multiunit activity as the electrode was advanced.
Microelectrode recording and stimulation included both foveal and
perifoveal striate cortex, whereas macroelectrode sites were restricted
to foveal cortex, as judged by published retinotopic maps (Daniel and
Whitteridge 1961; Tootell et al. 1982; Van Essen et al. 1984).
able from FHC) was added, which shunted the stimulating and return
electrodes between each stimulus pulse, thereby dissipating the charge
accumulated by the electrode capacitance. As an additional precaution, a high-input impedance oscilloscope (10 M⍀) could be placed
momentarily across the electrode leads to determine whether the
electrode potential during stimulation exhibited signs of excessive
polarization during each pulse (Brummer and Turner 1975). Because
connecting the oscilloscope in this way compromised the electrical
isolation of the stimulator, it was used only momentarily and only
when stimulus currents were sufficiently high to suspect that hazardous polarization might occur within the duration of a single pulse.
Measurement of electrode impedance was accomplished using a
1-kHz impedance meter (Bak IMP-1) that could be connected to the
electrode in place of the pulse generator. According to the manufacturer, sinusoidal current passed during a measurement was about 0.2
␮A root mean square (rms).
A standard AC coupled recording amplifier and high-impedance
headstage (Bak MDA-4) could also be connected to the electrode to
record the amplitude of the multiunit activity or flash-evoked potentials. To record background multiunit activity, the signal was bandpass filtered with half-amplitude at 200 Hz and 5 kHz. An index of the
amplitude of background activity was computed by digitally sampling
the signal at 16 kHz for 100 ms and then computing the SD of the
samples. (This measure is effectively proportional to the rms amplitude of the signal.) Field potentials were recorded in response to
flashes generated by a xenon strobe light shining on the back of an
opalescent viewing screen. In this case the band-pass filter limits were
set at 2 Hz and 2 kHz and the evoked response was averaged for as
many as 50 flashes.
MICROSTIMULATION OF STRIATE CORTEX
3445
RESULTS
Physiological identification of laminar position
FIG. 2. Relationship between flash-evoked potential and transition in background activity within striate cortex. A–E: each column illustrates potentials
evoked at successive depths along 5 microelectrode penetrations. Filled dots:
first appearance in each penetration of an early positive wave (positive
upward), corresponding to inversion of N1 component of evoked potential as
recorded from cortical surface (Snyder et al. 1979). Filled triangles: point at
which large drop in background activity recorded. Typically, appearance of
positive wave and drop in activity coincide closely (A–D), but occasionally
there was a discrepancy (E). Cortical depth between successively recorded
evoked potentials varies for each column: A, B, E: 100 ␮m; C: 160 ␮m, D: 80
␮m. Vertical bar at beginning of first trace in each column, 50 ␮V. Identical
timescale for all columns. Xenon strobe flash occurred at beginning of each
trace.
position were pooled. The laminar boundaries marked along
the x-axis demarcate the average width for each lamina as
computed from the pooled sample. A statistical 95% confidence interval for this curve was computed from the SE of each
point on the solid curve and is shown by the stippled region. As
can be seen (Fig. 1), background activity begins to increase as
the electrode approaches and enters layer 4, reaching a maximum within layer 4C, then falls abruptly as the electrode
passes out of layer 4 and into layers 5 and 6.
As a further aid in establishing cortical depths, the N1 and
P3 components of the flash-evoked potential (Snyder et al.
1979) reversed sign at or near the 4/5 border. Figure 2 shows
examples of evoked potentials recorded at regular intervals
along five penetrations through perifoveal striate cortex. In all
cases but E, the N1 inversion occurs just at or below the point
of abrupt diminution of background activity (Fig. 1). Although
this measure was not as consistently accurate as the transition
in spontaneous activity, it also provided useful corroboration
during the experimental sessions.
Laminar profiles of detection threshold
FIG. 1. Top: sketches of electrolytic lesions identifying transition from
high- to low-amplitude background activity. Major laminar subdivisions at left,
according to Lund (1973). Lesions A–C were placed at depth where drop in
activity was first distinguishable; lesions D–O were made at various depths
beyond this transition point, and the tops of the “brackets” indicate estimated
position of transition as judged from microdrive readings. Most discrepant
estimates (L, O), farthest from the transition, are likely to have the most error.
Bottom: average background activity profile for 15 penetrations through striate
cortex. Abscissa: cortical depth relative to point of transition in activity (at “0”)
in relation to boundary between layers 4 and 5. Major laminar subdivisions, as
judged from Nissl-stained sections, indicated above abscissa. Stippled zone:
95% confidence interval based on SE. Sharp drop in background activity
between layers 4 and 5 provides useful “on-line” marker of laminar position.
J Neurophysiol • VOL
Figure 3 summarizes a typical traverse in which the detection threshold and background activity were recorded at
100-␮m intervals. The primary reference point for aligning
tracks with the laminar pattern is shown as a solid triangle in
Fig. 3C, where the sudden drop in background activity marks
the layer 4/5 boundary (at L in Fig. 3A). Note that this activity
profile (Fig. 3C) is very similar to the average curve shown in
Fig. 1. The corresponding detection threshold profile (Fig. 3B)
displays characteristics typical of many penetrations (Fig. 4).
Below layer 1 there is a broad region of quite uniform low
threshold, followed by a sizable increase in layer 4C (Fig. 3B,
lesions K and L). Thresholds are again low in layer 5 but then
rise as the track enters layer 6. The star (Fig. 3B) indicates a
threshold that was ⬎25 ␮A, the maximum tested.
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
As noted by Vaughan (1982), the level of background
activity and the polarity of photically evoked potentials provide
“on-line” clues as to the laminar location of a microelectrode
within striate cortex. Such assessment was particularly useful,
and proved to be reasonably accurate. To assess the reliability
of changes in background activity in relation to laminar location it was measured in 15 penetrations in which identifying
lesions were made (Fig. 1). As the microeletrode penetrates
deeper into layer 4 there is a large, but gradual, buildup in
activity, and then within a few micrometers a very abrupt drop.
This drop corresponded closely to the border of layers 4 and 5,
as determined by the location of the identifying lesions and the
readings from the microdrive. The greatest discrepancies (Fig.
1, lesions L and O) were associated with two of the deeper
lesions that, consequently, are likely to be the least accurate.
To obtain a complete laminar profile of background activity,
data were pooled from 15 penetrations through all or most of
cortex (Fig. 1). Measured distances along each track were
adjusted for differences in individual laminar thickness and
overall cortical depth, and adjusted for histological shrinkage
(about 15%). Because corresponding sites were rarely in exactly the same laminar position, sites within 50 ␮m of a given
3446
E. A. DEYOE, J. D. LEWINE, AND R. W. DOTY
Threshold profiles from most penetrations shared the general
features illustrated in Fig. 3, although there were significant
variations. Detailed profiles spanning most of the cortical
layers were obtained for 16 penetrations. The associated background activity records, evoked potentials, and lesions were
used to align these penetrations with each other and with the
laminar architecture. Figure 4 shows six such threshold profiles, chosen to illustrate the maximum degree of variation
observed among the individual penetrations. Because each
penetration had a somewhat different average threshold, the
stimulus currents are expressed as a percentage deviation from
the mean for each penetration. Cortical depths are expressed
relative to the depth of the 4/5 border (marked by filled
triangles). The mean threshold current for each profile is
indicated by a dashed line (Fig. 4). The thresholds of Fig. 3B
are shown in normalized form in Fig. 4A. Profiles of background activity and electrode impedance were similar for each
penetration and have thus been omitted.
The first three variations, A, B, and C in Fig. 4, depict
common threshold profiles, characterized by an extended
course of low thresholds in the supragranular layers. Occasionally, as in Fig. 4B, this was preceded by a zone of sharply
elevated thresholds as the electrode passed through the meninges to enter the brain, possibly reflecting the consequences of
pressure. Usually, however, the electrode was advanced rapidly to penetrate the dura mater, and this high initial threshold
was not observed.
In almost all penetrations, the supragranular region of low
thresholds contained the point of greatest sensitivity (lowest
threshold) within the penetration. In contrast, a region of high
threshold was routinely encountered just superficial to the layer
4/5 boundary (Fig. 4, filled triangles on abscissa). Within
layers 2– 6, the highest thresholds were usually encountered
within layer 4C or at the end of the penetration (Fig. 4, C–F).
Another consistent feature was the low-threshold “notch” just
below the 4/5 boundary. Although most penetrations shared
J Neurophysiol • VOL
FIG. 4. Examples of laminar profiles for detection threshold in 6 penetrations, selected to show maximum range of variation encountered. In all panels
x-axes are identical and indicate cortical depth normalized relative to transition
in background activity, corresponding to layer 4/5 border (solid triangles).
Detection thresholds expressed as percentage deviation from the mean (dotted
line) for each penetration. Actual values of mean thresholds for each penetration are (in ␮A): 9.2 (A); 4.6 (B); 5.2 (C); 5.3 (D); 6.3 (E); 5.7 (F). All
ordinates have same scale except as indicated for D. Threshold values that
were too high to measure without risk of electrolytic damage indicated by
small stars. Penetrations A and B, ended just within layer 6, thus fail to show
high-threshold zone seen in penetrations C–F.
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
FIG. 3. Reconstruction of track through striate cortex (A) in which detection threshold (B) and background activity (C) recorded at successive sites in
layers 1– 6. Stimulation/recording sites in A indicated by letters associated with
small cross-bars along track. These letters also mark corresponding data points
in B and C. Reconstructed track aligned with cortical lamination using
transition in background activity, according to Figs. 1 and 2, occurring between
sites L and M, as a marker for the 4/5 border. Major laminar divisions
according to Lund (1973). B: corresponding profile of detection threshold.
Abscissa indicates depth, estimated from microdrive readings; identical scale
in B and C. Ordinate, peak amplitude of stimulus pulses at detection threshold.
Threshold at point marked by star is too high to measure without risk of
damage to cortex. C: amplitude of background activity recorded at same points
as in B. Transition from high to low activity marking layer 4/5 border indicated
by filled triangle.
these general features, four of 16 profiles displayed more
complex forms having secondary peaks (Fig. 4, D and E). Also
in four of 16 penetrations low thresholds persisted throughout
layers 3, 4, and 5 (e.g., Fig. 4F), but in all cases the lowest
thresholds occurred in the superficial layers whereas the highest thresholds were in deeper laminae.
The average threshold for each penetration tended to be
more similar within individual animals than between animals.
For one monkey, thresholds near 10 ␮A were the rule, whereas
our best animal had thresholds routinely ⬍5 ␮A. The lowest
individual threshold recorded was approximately 1 ␮A (Fig. 6,
lesion J). Equipment limited testing with still lower currents.
The ratio of highest to lowest threshold within each penetration
ranged from ⬎20 (22 vs. 1 ␮A, penetration D in Fig. 6) to ⬍2
(7.5 vs. 3.8 ␮A in penetration C; Fig. 6). Even higher ratios
may have existed but could not be measured because the
maximum test current was usually limited to 25 ␮A to avoid
hydrolytic damage.
Threshold data from 16 penetrations that sampled the full
depth of the cortex were combined to generate an average
threshold profile (Fig. 5). To accomplish this, all penetrations
were converted to the form used in Fig. 4, where depth
measurements were converted to normalized form and thresholds were expressed as percentage deviation from the mean for
each penetration. The data at corresponding laminar positions
(within 50-␮m bins) were then averaged. The 95% confidence
MICROSTIMULATION OF STRIATE CORTEX
interval based on the SE is indicated by the stippled region
(Fig. 5).
This average curve reflects the more common features of the
individual penetrations shown in Fig. 4. The curve (Fig. 5) is
distinguished by two regions of low threshold. The superficial
region is broad and includes layers 3, 4A, and at least part of
4B, whereas the infragranular region is confined mainly to the
upper portion of layer 5. Higher thresholds were found near the
surface, in layer 4C and in layer 6.
The true extent of the superficial zone of low thresholds may
not be accurately depicted by Fig. 5 because compression of
the dura mater and consequent depth inaccuracies were largest
in the most superficial layers. To circumvent this problem, 13
additional penetrations (Fig. 6) were undertaken specifically to
place marking lesions within the superficial low threshold
zone. In each of these cases the experiment was terminated
after the lesion because thresholds were then elevated in the
surrounding tissue. For Fig. 6 lesions A–G were placed at sites
where the detection threshold first dropped below 10 ␮A in
each penetration. From other complete penetrations, the 10-␮A
criterion was found to provide an accurate estimate of the
upper and lower margins of the sensitive region. Based on the
positions of the centers of these lesions, the upper boundary
was no more superficial than the middle of layers 2/3 (Fig. 6,
lesion F). Note that this more accurate estimate suggests that
the average threshold profile should probably rise more steeply
in layers 1 and 2 than is depicted in Fig. 5 where the upper
margin of the superficial zone appears to extend almost to layer
1, but the data of Fig. 6 indicate that this is not the case.
To identify the most sensitive positions within the cortex, a
few sites having very low thresholds were marked with lesions
(Fig. 6, lesions H–J). Although the thresholds at these sites
were notably low, 4.5, 2.7, and 1.0 ␮A, respectively, the
lesions at these points made it impossible to know whether
even lower thresholds could have been encountered had each
penetration continued. On the other hand, the 1.0-␮A threshold
(Fig. 6, lesion J) was as low as ever measured accurately.
Consequently, its position identifies layer 3 as the layer of
greatest sensitivity in the present experiments.
J Neurophysiol • VOL
In three more penetrations (Fig. 6, K–M), lesions were
placed at sites where the threshold fell below 10 ␮A during
retraction of an electrode initially advanced into the lower
cortical layers. These lesions indicated that the superficial zone
of low thresholds has a lower limit that is no deeper than layer
4B. (Control experiments in which threshold profiles were
compared during both insertion and retraction of the electrode
indicated virtually no alteration of thresholds in superficial
layers by passage of the electrode.) Taken together, all the
lesions in Fig. 6 indicate that a particularly sensitive region of
cortex having the lowest thresholds includes layers 3, 4A, and
at least part of 4B.
Thresholds were not routinely measured in white matter
because it was usually encountered late in an experimental
session if at all. On three occasions, however, the experiment
continued long enough to traverse the white matter. Generally,
thresholds were found to be significantly higher (two- to
fourfold) than in gray matter but the variability was large and
occasional low thresholds (e.g., 2.2 ␮A) were also observed. In
one penetration, the electrode passed from opercular cortex,
through white matter, and into calcarine cortex. On entering
calcarine gray matter the thresholds became lower and more
consistent. This suggested that the threshold variability in
white matter was not an artifact but rather reflected a true
heterogeneity. Altogether, calcarine V1 was encountered on
three occasions and in each case thresholds were found to be
comparable in magnitude to those obtained in opercular cortex.
Thus there were no obvious differences between stimulation of
peripheral versus central visual field representations.
DISCUSSION
The tantalizing question is not what do the macaques see,
but how the intrusion of electrical stimulation into the exquisite
neuronal complexity of the striate cortex consistently yields a
perceptual effect that is uniformly recognizable despite variation in (micro) locus or stimulus parameters. The evidence for
this putative uniformity is that, once trained to respond to
stimulation at one striate location, a macaque responds almost
unfailingly to stimulation of another locus, so long as it is in
striate cortex (Bartlett et al. 2005; Doty 1965, 1969; Doty and
Negrão 1973; Doty et al. 1973), as consistently found herein.
Furthermore, Tehovnik et al. (2005) and Bradley et al. (2005)
find that macaques, trained to saccade to 0.1– 0.2° lights
flashed at random loci, also make saccades to positions corresponding to the visual field location of stimulated points in
striate cortex, the presumption being that the animal experiences the stimulation as a localized phosphene. All this, of
course, is consonant with the human experience of the phos-
FIG. 6. Lesions marking limits of zone of low thresholds in supragranular
layers: A–G placed where threshold first dropped below 10 ␮A; H–J mark
sites of especially low threshold, 4.5, 2.7, and 1 ␮A, respectively, the latter
being the lowest recorded; K–M were placed at positions where threshold fell
below 10 ␮A during retraction of an electrode that was initially advanced into
layer 4C.
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
FIG. 5. Laminar profile of average detection threshold for 16 penetrations.
Abscissa: cortical depth normalized relative to border between layers 4 and 5,
according to Figs. 1 and 2. Major laminar divisions according to Lund (1973)
above this axis. Each point (solid line) is average percentage deviation in
threshold at that depth, whereas stippled area corresponds to 95% confidence
interval based on SE. Not all penetrations traversed fully through the subgranular layers, thus contributing, in part, to larger SEs observed in layer 6.
3447
3448
E. A. DEYOE, J. D. LEWINE, AND R. W. DOTY
J Neurophysiol • VOL
on average, about 34 myelinated fibers, mostly ⬍1.0 ␮m, as
well as dendrites and unmyelinated fibers, roughly 2,000 such
bundles/mm2. They are clearly associated with the columnar
organization of cortical circuitry.
The choice is thus most likely to be between, or attributable
to both, the bands of Baillarger and the vertical bundles of
myelinated fibers, mostly emanating from the pyramidal cells
of layer 2/3. The latter fibers might be good candidates, until it
is recalled that the densest gathering of these vertical bundles
occurs in layer 6 (see Fig. 19 in Peters and Sethares 1996),
where thresholds are distinctly high. On the other hand, although the lowest thresholds do seem to impinge on the stria of
Gennari in layers 4B and upper 4C␣ (Figs. 3; 4, A, B, and C;
and 6, lesions A, B, D, E, H, I, K, L, and M), there are
exceptions, and there is an impressive stretch of low thresholds
throughout layer 2/3. Some encouragement to the participation
of the horizontal bands is seen in the abrupt fall in threshold as
the electrode enters the less-dense inner band of Baillarger
(layer 5); and, commensurate with the lower fiber density, the
threshold there is higher than that in supragranular layers (Figs.
3 and 5).
Another logical candidate for higher excitability is the magnocellular system, the afferent input of which lies primarily
within layer 4C␣, just below the stria of Gennari and the
groups of small Meynert cells (Peters and Sethares 1991). The
recipient cells of the magnocellular input project into layers 4A
and 3, where they intermingle with afferents of the blue/
yellow, parvocellular system (Chatterjee and Callaway 2003;
Yabuta and Callaway 1998). Both are thus in ready range of
the lowest threshold loci. The magnocellular afferent terminals
cover sixfold as great an area as the parvocellular afferents
(Fitzpatrick et al. 1985), thus presenting a more diffuse target.
Furthermore the axons to the magnocellular system have a
relatively large diameter, thus greater excitability, and convey
a noncolor, luminance signal appropriate to phosphene production. However, the magnocellular system is essentially absent
at the fovea, and at 15° eccentricity is sevenfold more prevalent
than parafoveally (Assopardi et al. 1999), yet there is no
corresponding difference in threshold for striate loci representing parafoveal versus peripheral fields. In other words, were
the magnocellular system to play a major role in phosphene
initiation, low-threshold points should be found more frequently in calcarine rather than parafoveal areas, and this is not
the case; there appears to be no difference in threshold in
relation to location in the visual field.
Schmidt et al. (1996) in humans, Bartlett and Doty (1980),
Tehovnik et al. (2005), and the present study all found thresholds of 2 ␮A or lower with microelectrodes in striate cortex;
yet most thresholds are considerably above this (Figs. 3, 4, and
6). It must thus be concluded that certain points of exquisite
sensitivity exist. Because they are not reliably encountered by
sampling in the vertical dimension, there must be some special
columnar system linked to detectability that is infrequently
traversed; but can be excited from some distance at slightly
higher thresholds, particularly by the horizontal bands. Tangentially directed tracks might reveal such an organization.
Nevertheless, even if such relatively sparse columns exist, a
puzzle remains as to why the high background activity of the
parvocellular recipient layer, 4C␤, is consistently associated
with a higher threshold. This is all the more perplexing in that
layer 4C␤ sends five times as many synapses to layers 2/3 as
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
phene. It, too, has a consistent uniformity across a sizable scale
of stimulus parameters and is similar from one striate locus to
another. As noted previously (Bartlett et al. 2005), the congruence between macaque and human visual systems, from anatomy to psychophysics, strongly suggests that the macaque
experience is highly similar to that of humans, and that the
“phosphene circuit,” for want of a better term, has a peculiar
prevalence. In other words, it must be that essentially the same
type of elements and processes are engaged regardless of
where in striate cortex the stimulus is applied, and that the
differences in threshold noted herein simply represent the
distance at which the electrode lies relative to the critical
elements. Location is paramount only insofar as the phosphene
acquires a subjective spatial projection relevant to the site of
stimulation. Angular form is absent in human reports despite
its prevalence in another nonvisual intrusion into the striatal
network, the flickering geometrical patterns of the scintillating
scotoma (Grüsser 1995). Transcranial magnetic stimulation, on
the other hand, elicits the familiar phosphene (Fernandez et al.
2002).
It must be appreciated that the precisely ordered tangle of
fibers and neurons among which the microelectrode passes
would seem to offer numerous possibilities for exciting a wide
variety of effects (see, e.g., Douglas and Martin 2004); yet
either they do not, or the animal ignores them and awaits the
consistency of the putative phosphene. This is probably unknowable; but, given the fastidiousness with which macaques
persist in adhering to a successful modus operandi, one should
hesitate to predict that they confidently respond to pink lines on
one trial and blue stars on another as the electrode passes from
one to another assortment of stimulated elements. If, indeed,
the microstimulation does evoke a wide variety of visual
effects, this will seriously complicate prosthetic application of
the procedure, the assumption hitherto being that a uniformity
of effect allows ready presentation of patterns formed by
stimulation at multiple loci.
It bears note that although macaques readily respond to the
projected spatial position associated with electrical stimulation
of the corresponding anatomical loci in striate cortex or lateral
geniculate nucleus (Bradley et al. 2005; Pezaris and Reid 2004;
Tehovnik et al. 2005), they fail to find equivalence between
geniculate and striate stimulation (Bartlett et al. 2005; Schuckman et al. 1970). Were the animals simply responding to any
form of visual perturbation, at various loci in the visual field,
then the geniculate and striate effects, each having a projection
into the visual field, should be equated. Because they are not,
it implies that the animals are not responding merely to
detection of some visual effect at whatever locus it may appear,
but rather respond to some quality, putatively uniform throughout striate cortex, that differs significantly from that produced
by electrical stimulation of the lateral geniculate nucleus.
Thus the most useful clue, and assumption, presently available as to the nature of the underlying neuronal processes is the
variation in threshold with laminar position. If the phosphene
phenomenon is, as it seems, prepotent, the expectation is that
the most excitable elements must produce it. These would
likely be the small myelinated fibers revealed by electron
microscopy in horizontal bands, the inner and outer stripes of
Baillarger (the outer being the stria of Gennari), and the
vertical bundles of axons extending throughout the gray matter
(Peters and Sethares 1996). These vertical bundles incorporate,
MICROSTIMULATION OF STRIATE CORTEX
J Neurophysiol • VOL
the retinal population. Yet they have an exceedingly wide
distribution of dendrites and, given that the geniculate input to
striate cortex is a mere 6 – 8% (Kara and Reid 2003; Latawiec
et al. 2000; Peters et al. 1994), to provide the full gamut of
visual experience, the giant RGCs might well feed a cortical
photergic circuit sufficient to yield the simplicity of a phosphene, and likewise to support such an effect from electrical
stimulation of the retina. Prima facie, this is contradictory to
the observation that threshold for detection of electrical stimulation in striate cortex is independent of illumination (Bartlett
et al. 2005). The electrical stimulation, however, may well be
prepotent over the less-vigorous discharge elicited by normal
visual input.
Such speculation, however, fails to provide insight into the
laminar profiles of detectability. Explaining the isolated points
of high excitability, together with the wide variation from one
penetration or locus to another, will probably remain a perplexing question for some time. The microstimulation is, perforce, being applied at random within an intricate and highly
organized network, with multiple functional nodes, and the
simplicity of the “phosphene output,” if such it be, need bear
no direct relation to the pathways by which it is accessed.
ACKNOWLEDGMENTS
Present addresses: E. A. DeYoe, Dept. of Radiology, Medical College of
Wisconsin, 8701 Watertown Plank Road, Milwaukee, WI 53226; J. D. Lewine,
Department of Neurology, MEG Program, Hoglund Brain Imaging Center,
University of Kansas Medical School, Kansas City, KS 66160.
REFERENCES
Azzopardi P, Jones KE, and Cowey A. Uneven mapping of magnocellular
and parvocellular projections from the lateral geniculate nucleus to the
striate cortex in the macaque monkey. Vision Res 39: 2179 –2189, 1999.
Bak M, Girvin JP, Hambrecht FT, Kufta CV, Loeb GE, and Schmidt EM.
Visual sensations produced by intracortical microstimulation of the human
occipital cortex. Med Biol Eng Comput 28: 257–259, 1990.
Bartlett JR, DeYoe EA, Doty RW, Lee BB, Lewine JD, Negrão N and
Overman WH Jr. Psychophysics of electrical stimulation of striate cortex
in macaques. J Neurophysiol 94: 3430 –3442, 2005.
Bartlett JR and Doty RW. An exploration of the ability of macaques to detect
microstimulation of striate cortex. Acta Neurobiol Exp (Warsz) 40: 713–728,
1980.
Bradley DC, Troyk PR, Berg JA, Bak M, Cogan S, Erickson R, Kufta C,
Mascaro M, McCreery D, Schmidt EM, Towle VL, and Xu H. Visuotopic mapping through a multichannel stimulating implant in primate V1.
J Neurophysiol 93: 1659 –1670, 2005.
Brindley GS. Report to the conference on visual prosthesis. In: Visual
Prosthesis, The Interdisciplinary Dialogue, edited by Sterling TD. New
York: Academic Press, 1971, p. 41– 48.
Brindley GS and Lewin WS. The sensations produced by electrical stimulation of the visual cortex. J Physiol 196: 479 – 493, 1968.
Brindley GS and Rushton DN. Implanted stimulators of the visual cortex as
visual prosthetic devices. Trans Am Acad Ophthalmol Otolaryngol 78: 741,
1974.
Brummer SB and Turner MJ. Electrical stimulation of the nervous system:
the principle of safe charge injections with noble metal electrodes. Bioelectrochem Bioenerg 2: 13–25, 1975.
Chatterjee S and Callaway EM. Parallel colour-opponent pathways to
primary visual cortex. Nature 426: 668 – 671, 2003.
Dacey DM, Liao HW, Peterson BB, Robinson FR, Smith VC, Pokorny J,
Yau KW, and Gamlin PD. Melanopsin-expressing ganglion cells in primate retina signal colour and irradiance and project to the LGN. Nature 433:
749 –754, 2005.
Daniel PM and Whitteridge D. The representation of the visual field on the
cerebral cortex in monkeys. J Physiol 159: 203–221, 1961.
Dobelle WH and Mladejovsky MG. Phosphenes produced by electrical
stimulation of human occipital cortex, and their application to the development of a prosthesis for the blind. J Physiol 243: 553–576, 1974.
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
does layer 4C␣ (Yabuta and Callaway 1998). Because there is
no interaction between threshold and level or type of illumination (Bartlett et al. 2005), the higher threshold seems unlikely to arise from some form of surround inhibition. Refractoriness from the background activity, and the lower excitability of the smaller, “parvocellular” system might be called on
for explanation.
A geniculate “loop” is not indicated, either from antidromic
stimulation or orthodromically by excitation of the corticothalamic system of layer 6, to produce an inhibitory effect, to
account for the high thresholds of that lamina. Similarly, it is
hard to imagine that antidromic effects into circumstriate areas
are significantly involved, although orthodromic transmission
out of striate cortex is obviously required to produce any
evidence of detection.
Finally, there is the puzzle that electrical, or mechanical,
stimulation of the retina in human subjects also produces a
“phosphene” (e.g., Humayun et al. 2003). Indeed, the very
word phosphene originated with the Greeks, in the clever idea
that the eye itself produced the light required for seeing
(Grüsser and Hagner 1990). The question then is whether the
retinally produced phosphene is basically the same as that
produced by electrical or transcranial magnetic stimulation of
striate cortex. Preliminary data (Weiland et al. 1999), from
local electrical stimulation of the retina, are not typical of
cortically elicited phosphenes. Were there a high degree of
similarity between retinally and cortically elicited phosphenes,
there should be an “inherent” cortical circuit to decode the
retinal phosphene-producing input. In such a case the cortically
elicited phosphene might result from activation of this inherent
pathway, rather than as a mere perceptual accompaniment of
the bizarre simultaneous activation of a hodgepodge of cortical
elements otherwise subserving normal visual processing. Without a careful comparison in man of the qualitative characteristics of the two (?) types of phosphene, retinally versus
cortically produced, the question is moot.
However, the basis for a possibly productive speculation has
just emerged. Dacey et al. (2005) recently described in macaques giant retinal ganglion cells (RGCs), containing melanopsin, that are inherently photically responsive and, unlike
their congeners in rodents, project to the lateral geniculate
nucleus. In addition to their inherent responsiveness, these cells
also convey a broad range of rod and cone inputs. Overall, their
characteristics suggest a luminance channel, described as the
luxotonic units by Kayama et al. (1979) in macaque striate
cortex. In the unanesthetized macaque some 20% of the units
in striate cortex respond for several minutes to steady, diffuse,
featureless illumination, the “photergic” variety of luxotonic
activity (as opposed to 8%, the “scotergic” units, that respond
correspondingly to darkness). Such photergic units are clearly
relevant to phosphenes, but most also respond to other types of
visual input, i.e., they are “simple” and “complex” cells as well
(Riso et al. 1979). From the observations of Dacey et al. (2005)
the mixed receptor input plus their melanopsin signal suggest
that the giant RGCs provide a reasonable basis for the characteristics of the photergic units of striate cortex. Furthermore,
the giant size of these cells is concordant with their being the
most electrically excitable elements of the retina and, by
extrapolation, possibly maintaining that size and excitability
advantage into the cortex as well. The major constraint in this
argument is the rarity of the giant RGCs, only about 0.2% of
3449
3450
E. A. DEYOE, J. D. LEWINE, AND R. W. DOTY
J Neurophysiol • VOL
Peters A, Payne BR, and Budd J. A numerical analysis of the geniculocortical input to striate cortex in the monkey. Cereb Cortex 4: 215–229, 1994.
Peters A and Sethares C. Organization of pyramidal neurons in area 17 of
monkey visual cortex. J Comp Neurol 306: 1–23, 1991.
Peters A and Sethares C. Myelinated axons and the pyramidal cell modules
in monkey primary visual cortex. J Comp Neurol 365: 232–255, 1996.
Pezaris JS and Reid RC. Microstimulation in LGN produces focal visual
percepts. Soc Neurosci Abstr 598.5, 2004.
Pollen DA. Some perceptual effects of electrical stimulation of the visual
cortex in man. In: The Nervous System. The Clinical Neurosciences, edited
by Tower DM and Chase TN. New York: Raven Press, 1975, vol. 2, p.
519 –528.
Ranck JB Jr. Which elements are excited in electrical stimulation of mammalian central nervous system: a review. Brain Res 98: 417– 440, 1975.
Riso RR, Brust-Carmona H, Bartlett JR, and Doty RW. Receptive field
properties of luxotonic units in macaque striate cortex. Soc Neurosci Abstr
5: 805, 1979.
Rushton DN and Brindley GS. Properties of cortical electrical phosphenes.
In: Frontiers in Visual Science, edited by Cool SJ and Smith EL. New York:
Springer-Verlag, 1978, p. 574 –593.
Schmidt EM, Bak MJ, Hambrecht FT, Kufta CV, O’Rourke DK, and
Vallabhanath P. Feasibility of a visual prosthesis for the blind based on
intracortical microstimulation of the visual cortex. Brain 119: 507–522,
1996.
Schuckman H, Kluger A, and Frumkes TE. Stimulus generalization within
the geniculo-striate system of the monkey. J Comp Physiol Psychol 73:
494 –500, 1970.
Snyder EW, Beck EC, and Dustman RE. Visual evoked potentials in
monkeys. EEG Clin Neurophysiol 47: 430 – 440, 1979.
Tehovnik EJ, Slocum WM, Carvey CE, and Schiller PH. Phosphene
induction and the generation of saccadic eye movements by striate cortex.
J Neurophysiol 93: 1–19, 2005.
Tootell RBH, Silverman MS, Switkes E, and DeValois RL. Deoxyglucose
analysis of retinotopic organization in primate striate cortex. Science 218:
902–904, 1982.
Van Essen DC, Newsome WT, and Maunsell JHR. The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual variability. Vision Res 24: 429 – 448, 1984.
Vaughan HG. The neural origins of human event-related potentials. Ann NY
Acad Sci 388: 125–138, 1982.
Weiland JD, Humayun MS, Dagnelie G, de Juan E Jr, Greenberg RJ, and
Iliff NT. Understanding the origin of visual percepts elicited by electrical
stimulation of the human retina. Graef Arch Klin Exp Ophthalmol 237:
1007–1013, 1999.
Yabuta NH and Callaway EM. Functional streams and local connections of
layer 4C neurons in primary visual cortex of the macaque monkey. J Neurosci 18: 9489 –9499, 1998.
94 • NOVEMBER 2005 •
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.33.3 on June 17, 2017
Doty RW. Conditioned reflexes elicited by electrical stimulation of the brain
in macaques. J Neurophysiol 28: 623– 640, 1965.
Doty RW. Electrical stimulation of the brain in behavioral context. Ann Rev
Psychol 20: 289 –320, 1969.
Doty RW and Bartlett JR. Stimulation of the brain with metallic electrodes.
In: Electrical Stimulation Research Techniques, edited by Patterson MM
and Kesner RP. New York: Academic Press, 1981, p. 71–103.
Doty RW and Negrão N. Forebrain commissures and vision. In: Handbook of
Sensory Physiology: Central Processing of Visual Information, edited by
Jung R. Berlin: Springer-Verlag, 1973, pt. B, p. 543–582.
Doty RW, Negrão N, and Yamaga K. The unilateral engram. Acta Neurobiol
Exp (Warsz) 33: 711–728, 1973.
Douglas RJ and Martin KAC. Neuronal circuits of the neocortex. Ann Rev
Neurosci 27: 419 – 451, 2004.
Evarts EV. A technique for recording activity of subcortical neurons in
moving animals. EEG Clin Neurophysiol 24: 83– 86, 1968.
Fernandez E, Alfaro A, Tormos JM, Climent R, Martinez M, Vilanova H,
Walsh V, and Pascual-Leone A. Mapping of the human visual cortex using
image-guided transcranial magnetic stimulation. Brain Res Protocols 10:
115–124, 2002.
Fitzpatrick D, Lund JS, and Blasdel GG. Intrinsic connections of macaque
striate cortex: afferent and efferent connections of lamina 4C. J Neurosci 5:
3329 –3349, 1985.
Grüsser OJ. Migraine phosphenes and the retino-cortical magnification factor.
Vision Res 35: 1125–1134, 1995.
Grüsser OJ and Hagner M. On the history of deformation phosphenes and
the idea of internal light generated in the eye for the purpose of vision. Doc
Ophthalmol 74: 57– 85, 1990.
Henderson DC, Evans JR, and Dobelle WH. The relationship between
stimulus parameters and phosphene threshold/brightness, during stimulation
of human visual cortex. Trans Am Soc Artif Intern Organs 25: 367–371,
1979.
Humayun MS, Weiland JD, Fujii GY, Greenberg R, Williamson R, Little
J, Mech B, Cimmarusti V, Van Boemel G, Dagnelie G, and de Juan E.
Visual perception in a blind subject with a chronic microelectronic retinal
prosthesis. Vision Res 43: 2573–2581, 2003.
Kara P and Reid RC. Efficacy of retinal spikes in driving cortical responses.
J Neurosci 23: 8547– 8557, 2003.
Kayama Y, Riso RR, Bartlett JR, and Doty RW. Luxotonic responses of
units in macaque striate cortex. J Neurophysiol 42: 1495–1517, 1979.
Latawiec D, Martin KA, and Meskenaite V. Termination of the geniculocortical projection in the striate cortex of macaque monkey: a quantitative
immunoelectron microscopic study. J Comp Neurol 419: 306 –319, 2000.
Lund JS. Organization of neurons in the visual cortex, area 17, of the monkey.
J Comp Neurol 147: 455– 496, 1973.
McNicol D. A Primer of Signal Detection Theory. London: Allen & Unwin,
1972.