Behavioural report of single neuron stimulation in
Arthur R. Houweling1,2& Michael Brecht1,2
Understanding how neural activity in sensory cortices relates to
perception isacentral themeofneuroscience. Actionpotentials of
sensory cortical neurons can be strongly correlated to properties
of sensory stimuli1and reflect the subjective judgements of an
individual about stimuli2. Microstimulation experiments have
established a direct link from sensory activity to behaviour3,4, sug-
gesting that small neuronal populations can influence sensory
decisions5. However, microstimulation does not allow identifica-
tion and quantification of the stimulated cellular elements6. The
sensory impact of individual cortical neurons therefore remains
unknown. Here we show that stimulation of single neurons in
somatosensory cortex affects behavioural responses in a detection
task. We trained rats to respond to microstimulation of barrel
cortex at low current intensities. We then initiated short trains
ofactionpotentials insingle neuronsbyjuxtacellularstimulation.
Animals responded significantly more often in single-cell stimu-
lation trials than in catch trials without stimulation. Stimulation
response was induced. Whereas stimulation of putative excitatory
neurons led to weak biases towards responding, stimulation of
putative inhibitory neurons led to more variable and stronger
sensory effects. Reaction times for single-cell stimulation were
long and variable. Our results demonstrate that single neuron
activity can cause a change in the animal’s detection behaviour,
suggesting a much sparser cortical code for sensations than pre-
Based on its volume7and density of neurons8, rat somatosensory
cortex contains an estimated two million neurons. The detection of
single-cell stimulation might therefore be a difficult task, and we
adopted a behavioural paradigm designed for observing single-cell
effects (Fig. 1a). We first trained animals to report short (200ms)
trains of microstimulation pulses. Stimulation of somatosensory
cortex evokes tactile sensations in humans9, and animal studies
have demonstrated an interchangeability of tactile stimuli and
cortical microstimulation10. We mainly targeted deep cortical
layers, where microstimulation detection thresholds are lowest in
rat barrel cortex11. Tongue lick responses were rewarded with a drop
of water and counted as a hit if a lick occurred within 100–1200ms
from stimulus onset (Fig. 1b). Animals typically learned this
microstimulation report task in a single session and detection
thresholds decreased to 2–5mA within days (Supplementary
Fig. 1). These values are comparable to the lowest cortical micro-
stimulation detection thresholds reported in humans12and ani-
mals11,13. To be able to detect potentially weak effects of single-cell
stimulation, we encouraged guessing (a non-conservative response
criterion) by introducing only mild negative reinforcement (a 1.5s
time-out) for false-positive responses (licks without preceding
Once animals responded consistently to low microstimulation
currents, weapproachedacorticalneuron closelywithaglasspipette
and evoked short (200ms) trains ofaction potentials byjuxtacellular
stimulation, a technique developed to label individual neurons14.
Juxtacellular stimulation currents (3–43nA, mean 12.6nA) strongly
On average, we evoked 14.266.5 (s.d.) action potentials during
current injection, a 25-fold increase over the average spontaneous
firing rate. The close apposition of neuron and pipette in the juxta-
cellular configuration in behaving animals typically resulted in short
neuron (at an average distance of about 75mm) was adjusted such
Center, PO Box 2040, 3000 CA, Rotterdam, The Netherlands.
AP rate (Hz)
Figure 1 | Behavioural setup and single-cell stimulation. a, Stimulation
experiments were performed in the barrel cortex of awake rats. Animals
responded to stimulation by interrupting a light beam (dashed line) with
and reward was delivered for correct responses (right). Top, single-cell
stimulation pipette with stimulation current wave form (upper) and
of stimulus were presented at random intervals (Poisson process, mean 3s):
microstimulation (2–8mA) (40% probability), juxtacellular single-cell
stimulation (40%) and no (or subthreshold) current injection ‘catch’ trials
to presentation of the next stimulus (left box) and were rewarded after a
juxtacellular current injection. Triangles indicate stimulation onset and
offset artefacts. d, Evoked action potentials (open circles) in a series of
stimulation trials. Spontaneous action potentials (solid circles) were
spontaneous and evoked action potentials; the right y axis label applies to
evoked action potentials.
Vol 451|3 January 2008|doi:10.1038/nature06447
resulting in an average hit (detection) rate of 75%. The action
potential firing of most cells was affected during and after micro-
stimulation (Supplementary Fig. 2).
Microstimulation andsingle-cell stimulation trialswere randomly
interleaved with ‘catch’ trials (with no or subthreshold current injec-
tion) (Fig. 1b). In paradigms with random stimulus presentation
times, catch trials can be used to estimate chance performance and
to guard against inadvertent cues15. To assess single-cell effects when
stimulation and catch trials flanked by correct microstimulation
Figure 2 shows an experiment on a regular spiking layer 5b pyr-
amidal neuron with a slender apical dendrite (Fig. 2a). Juxtacellular
stimulation evoked on average 9.1 action potentials during the
mainly after single-cell stimulation (Fig. 2b top) and micro-
stimulation (Fig. 2b bottom), but only once after no stimulation
catch trials (Fig. 2b middle). Quantification of responses (Fig. 2c)
suggests that the animal reported single pyramidal cell activity. Even
effects, this effect was not significant on the single neuron level
(P50.099, Fisher’s exact test). This is not unexpected given the
limited number of trials (see Methods).
A population analysis revealed, however, that single-cell stimu-
lation biased animals towards responding. Figure 3a shows, for
51 neurons, that animals responded significantly more often in
single-cell stimulation trials (mean hit rate 22.0%) than in no-
current-injection catch trials (mean false-positive rate 17.9%;
of the stimulated neuron rather than on inadvertent cues associated
with the current injection, we stimulated a further set of 19 neurons.
We applied single-cell stimulation as usual, but instead of the no-
stimulation current) catch trials. Subthreshold current injections
activated neurons only weakly or not at all. Animals also responded
significantly more often in single-cell stimulation trials than in
subthreshold catch trials (Fig. 3b; mean hit rate 27.4%; mean false-
positive rate 20.6%; P50.019). Stimulation effects were distributed
evenly across animals (Supplementary Fig. 3). Having verified that
single-cell stimulation led to significant biases in two independent
sets of neurons (Fig. 3a, b), we wanted to confirm that this effect did
notresultfromtheinadvertentstimulation ofneighbouring neurons
or other nonspecific effects. Thus, we injected current (25nA, twice
the average current applied with juxtacellular stimulation) through
the pipette into extracellular space (instead of applying it to a neu-
ron). These control experiments showed that animals did not report
current injection into extracellular space (Fig. 3c; n590; mean hit
rate 18.7%; mean false-positive rate 19.0%; P50.598). To test if
single-cell stimulation effects (Fig. 3a, b) were different from those
of extracellular current injection (Fig. 3c), we compared effect size
(hit rate2catch trial response rate) across those two data sets and
observed a significant difference (P50.008). Finally, we tested if
single-cell stimulation effects were specific to the attended (and
trained) cortical area. As before, microstimulation was applied to
barrel cortex, but we now stimulated single neurons in visual cortex.
Animals did not report single-cell stimulation in visual cortex
(Fig. 3d; n521; mean hit rate 21.2%; mean false-positive rate
20.1%; P50.319), suggesting that stimulation effects are specific
to the attended cortical area.
Further observations show that the animals’ responses were
caused by the stimulation of single and not multiple neurons.
(1) Juxtacellular stimulation currents were approximately three
motor or sensory responses with microstimulation (2–200mA).
(2) Although we occasionally observed the inadvertent stimulation
ofasecond neuronbytheappearanceofasecondlargeaction poten-
tial waveform in our recordings, such inadvertent stimulation was
experiments; Supplementary Fig. 4). All results presented here were
also significant when single-cell stimulation trials with secondary
action potentials were excluded. (3) Firing rates of more distant cells
(with action potentials less than 0.5mV) were not modulated
(Supplementary Fig. 5). (4) Juxtacellular labelling typically fills
Because microstimulation in barrel cortex can evoke whisker
movements, we combined stimulation experiments with whisker
tracking to assess if rats sense single-cell stimulation indirectly by
anism: near detection threshold microstimulation did not evoke
movements even though it was reported (Supplementary Fig. 6a)
and single-cell stimulation did not evoke whisker movements
(Supplementary Fig. 6b).
weak on average (approximate 5% effect size: single-cell stimulation
hit rate–catch trial response rate). As illustrated in Supplementary
Fig. 7, the strength of the effect depended greatly on the animal’s
Response No response
–101 Time (s)
Figure 2 | Behavioural responses to
stimulation of a single layer 5b
of the stimulated neuron with
dendritic tree (red) and axon (blue,
and a tungsten microstimulation
electrode aligned along the electrode
track. Barrel rows (brown) are
labelled with letters. L, layer; WM,
white matter. b, Action potential
(ticks) raster plots and first lick
responses (red squares) during
trials (top), no-current-injection
selected microstimulation trials
(bottom). The neuron was inhibited
during and after microstimulation
c, Quantification of responses to
single-cell stimulation, catch trials
NATURE|Vol 451|3 January 2008
however, the effect size was about 9% (and significant). Most inter-
estingly, the variance of stimulation-evoked sensory effects was
greater for putative interneurons than putative excitatory cells
(P50.0023, see Methods), which we distinguished based on spike
width and firing pattern (Supplementary Fig. 8). Whereas stimu-
lation of putative excitatory neurons led to weak but significant
biases towards responding, stimulation of putative inhibitory neu-
rons led to stronger and more variable sensory effects (Fig. 3e). In
than in any of the 59 putative excitatory cells or 90 control experi-
ments. In two of these three putative interneurons most hits were
observed for trials in which the evoked firing rates were not higher
than the population average, suggesting that interneuron action
tials of excitatory cells.
Reaction times for single-cell stimulation were long and variable
(Figs 2 and 4a, b) compared with microstimulation responses
(Fig. 4c). Although effect size could be considerable in individual
cells (Fig. 2), we did not observe single-cell responses with close to
100% hit rates and short reaction times as observed often in micro-
saturating signal. Microstimulation at 2–8mA presumably activates
multiple neurons, which may account for the difference between
single-cell stimulation and microstimulation. Even smaller currents
(less than 2mA) are known to activate cortical neurons16. It remains
to be seen if such currents lead to a microstimulation performance
comparable to that of single-cell stimulation.
The combination of single-cell stimulation and control experi-
ments shows that the activity of single sensory cortical neurons can
lead to a behaviourally reportable effect. It has been estimated that a
cells that generate about 1,550 spontaneous action potentials in a
200ms period and about 4,000 action potentials in response to a
small (3.3u) whisker deflection17that is close (about 60% hit rate)
to the animal’s detection threshold18. Given these numbers it is sur-
a neuron is detectable. Other measurements suggest lower rates of
False positives (%)
Hits – false positives (%)
False positives (%)
4060 800 100
False positives (%)
P = 0.319
P = 0.019
stimulation hits (%)
0 10205060 70
Extracellular stimulation hits (%)
Viual cortex single-cell
stimulation hits (%)
Single-cell stimulation hits (%)
Single-cell stimulation hits (%)
P = 0.598
P = 0.022
Figure 3 | Initiation of action potentials in single barrel cortex neurons
causes biases towards responding. a, Response rates for single-cell
stimulation trials (hits) versus no-current-injection catch trials (false
positives) (n551 neurons; note several points coincide). Fast spiking,
putative interneurons (filled circles); non-fast spiking, putative excitatory
neurons (empty circles). b, Response rates for single-cell stimulation trials
(hits) versus subthreshold current injection catch trials (n519 neurons).
Conventions as in a. c, Response rates for trials in which we applied 25nA
(twice the average juxtacellular stimulation current) into extracellular space
(hits) versus no-current-injection catch trials (false positives) (n590
stimulation sites). d, Response rates for single-cell stimulation trials (hits)
versus no-current-injection catch trials (false positives) for visual cortex
neurons (n521), while microstimulation was applied in barrel cortex.
Animals had been trained to report microstimulation in barrel cortex.
e, Distributionofsensoryeffects (single-cell stimulationhit rate2catch trial
response rate) across putative interneurons and putative excitatory neurons
in barrel cortex single-cell stimulation experiments (conventions as in a).
Microstimulation hits – FPs
cumulative difference (%)
Reaction time (s)
Cumulative responses (%)
Single-cell stimulation hits – FPs
cumulative difference (%)
Figure 4 | Reaction times for single-cell stimulation are long and variable
compared with microstimulation responses. a, Cumulative distribution of
reactiontimesfor microstimulation (dashed),single-cellstimulation(solid)
and catch trials (dotted). b, Difference of the cumulative distributions of
reaction times for single-cell stimulation and catch trials. This isolates the
contribution of single-cell stimulation from false-positive responses. The
vertical line marks the time where 50% of the peak difference is reached; the
grey area marks the time from 25% to 75% of the peak difference.
c, Difference of the cumulative distributions of reaction times for
microstimulation and catch trials. Conventions as in b.
NATURE|Vol 451|3 January 2008
ongoing19–21and sensory-evoked22–25cortical action potential acti-
vity. The detectability of single-cell stimulation might therefore be
related to the sparseness of cortical activity. A single cortical pyr-
amidal cell connects to several thousand postsynaptic neurons26,
but most of these connections are weak27. Likewise, a single inhibi-
tory neuron connects to thousands of local neurons. Depending on
ongoing membrane potential fluctuations, variable sets of post-
synaptic cells may become activated or suppressed, which might
contribute to the variable reaction times. Some models of sensory
decision making contain a temporal integration step during which
sensory evidence is accumulated28. It is conceivable that the weak
single-cell signals require longer temporal integration, thereby con-
tributing to the long reaction times. The mechanisms that hold the
sensory information between single-cell stimulation and reaction
remain to be determined.
ject locally) can lead to strong sensory effects suggests that (1) local
circuits are involved in the read-out of single-cell activity and that
tials. Cortical microstimulation evokes pronounced and long-lasting
have been trained to detect the suppression of activity. The present
results, with the classic single afferent stimulation experiments by
Vallbo and colleagues29and single-cell stimulation experiments in
rat motor cortex30, demonstrate the behavioural relevance of single
responses as in classical physiology. In further studies it should be
possible to establish how the frequency and number of action poten-
tials are related to the evoked sensations.
Animals were trained to report microstimulation (40 cathodal pulses at 200Hz,
0.3ms pulse duration) applied to the barrel cortex through a tungsten micro-
electrode at a depth of 1,500mm. In the first training session, current intensities
no greater than 200mA were applied; subsequently current intensity was
decreased according to detection performance. During training animals were
put on a water restriction schedule with daily access to water ad libitum for one
hour after the experiment. Once the animal performed at detection thresholds
no greater than 5mA in at least one block of trials on two consecutive days, we
switched to the single-cell stimulation report task; here microstimulation cur-
rents were on average 5.061.6mA. To stimulate single neurons close to the
microstimulation site, we glued a tungsten microelectrode close to the tip of a
glass pipette (average tip separation approximately 75mm). The construct was
actual depth from the cortical surface was less because of dimpling and oblique
penetrations. From the histologically identified neurons it appears that most
single-cell stimulation experiments were performed in cortical layers 4, 5A and
5B. Cells were classified as fast spiking neurons (putative interneurons) if the
responded with at least 50 action potentials (that is, at least 250Hz) during at
least one 200ms current injection (see Supplementary Fig. 8). The remainder of
cells were classified as non-fast spiking (putative excitatory) cells.
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 7 July; accepted 1 November 2007.
Published online 19 December 2007.
1. Rieke, F., Warland, D., de Ruyter Van Steveninck, R. R. & Bialek, W. Spikes:
Exploring the Neural Code (MIT Press, Cambridge, Massachusetts, 1997).
Britten, K. H., Newsome, W. T., Shadlen, M. N., Celebrini, S. & Movshon, J. A. A
relationship between behavioral choice and the visual responses of neurons in
macaque MT. Vis. Neurosci. 13, 87–100 (1996).
Afraz, S. R., Kiani, R. & Esteky, H. Microstimulation of inferotemporal cortex
influences face categorization. Nature 442, 692–695 (2006).
Salzman, C. D., Britten, K. H. & Newsome, W. T. Cortical microstimulation
influences perceptual judgements of motion direction. Nature 346, 174–177
5. Murasugi, C. M., Salzman, C. D. & Newsome, W. T. Microstimulation in visual
area MT: effects of varying pulse amplitude and frequency. J. Neurosci. 13,
Tehovnik, E. J. Electrical stimulation of neural tissue to evoke behavioral
responses. J. Neurosci. Methods 65, 1–17 (1996).
Zilles, K. & Wree, A. in The Rat Nervous System (ed. Paxinos, G.) 649–685
(Academic Press, London, 1995).
Beaulieu, C. Numerical data on neocortical neurons in adult rat, with special
reference to the GABA population. Brain Res. 609, 284–292 (1993).
Penfield, W. & Rasmussen, T. The Cerebral Cortex of Man: A Clinical Study of
Localization of Function (Macmillan, New York, 1950).
10. Romo, R., Hernandez, A., Zainos, A. & Salinas, E. Somatosensory discrimination
based on cortical microstimulation. Nature 392, 387–390 (1998).
12. Schmidt, E. M. et al. Feasibility of a visual prosthesis for the blind based on
intracortical microstimulation of the visual cortex. Brain 119, 507–522 (1996).
13. Bartlett, J. R. & Doty, R. W. An exploration of the ability of macaques to detect
microstimulation of striate cortex. Acta Neurobiol. Exp. (Wars.) 40, 713–727
14. Pinault, D. A novel single-cell staining procedure performed in vivo under
thalamic cells and other central neurons with biocytin or Neurobiotin. J. Neurosci.
Methods 65, 113–136 (1996).
15. Talwar, S. K. & Gerstein, G. L. Auditory frequency discrimination in the white rat.
Hear. Res. 126, 135–150 (1998).
16. Swadlow, H. A. Monitoring the excitability of neocortical efferent neurons to
direct activation by extracellular current pulses. J. Neurophysiol. 68, 605–619
17. de Kock, C. P., Bruno, R. M., Spors, H. & Sakmann, B. Layer- and cell-type-specific
suprathreshold stimulus representation in rat primary somatosensory cortex.
J. Physiol. (Lond.) 581, 139–154 (2007).
18. Stuttgen, M. C., Ruter, J. & Schwarz, C. Two psychophysical channels of whisker
deflection in rats align with two neuronal classes of primary afferents. J. Neurosci.
26, 7933–7941 (2006).
19. Kerr, J. N., Greenberg, D. & Helmchen, F. Imaging input and output of neocortical
networks in vivo. Proc. Natl Acad. Sci. USA 102, 14063–14068 (2005).
20. Margrie, T. W., Brecht, M. & Sakmann, B. In vivo, low-resistance, whole-cell
recordings from neurons in the anaesthetized and awake mammalian brain.
Pflugers Arch. 444, 491–498 (2002).
21. Waters, J. & Helmchen, F. Background synaptic activity is sparse in neocortex.
J. Neurosci. 26, 8267–8277 (2006).
22. Brecht, M. & Sakmann, B. Dynamic representation of whisker deflection by
layer 4 rat somatosensory cortex. J. Physiol. (Lond.) 543, 49–70 (2002).
23. Moore, C. I. & Nelson, S. B. Spatio-temporal subthreshold receptive fields in the
vibrissa representation of rat primary somatosensory cortex. J. Neurophysiol. 80,
24. Olshausen, B. A. & Field, D. J. Sparse coding of sensory inputs. Curr. Opin.
Neurobiol. 14, 481–487 (2004).
25. Warren, R. A. & Dykes, R. W. Population analysis of single neurons in cat
somatosensory cortex. Somatosens. Mot. Res. 9, 297–312 (1992).
26. DeFelipe, J. & Farinas, I. The pyramidal neuron of the cerebral cortex:
morphological and chemical characteristics of the synaptic inputs. Prog.
Neurobiol. 39, 563–607 (1992).
27. Song, S., Sjostrom, P.J., Reigl,M.,Nelson, S.&Chklovskii, D.B. Highlynonrandom
28. Gold, J. I. & Shadlen, M. N. Representation of a perceptual decision in developing
oculomotor commands. Nature 404, 390–394 (2000).
tactileafferents from thehuman hand.Sensory attributesrelatedtounittypeand
properties of receptive fields. Brain 107, 727–749 (1984).
30. Brecht, M., Schneider, M., Sakmann, B. & Margrie, T. W. Whisker movements
evoked by stimulation of single pyramidal cells in rat motor cortex. Nature 427,
Supplementary Information is linked to the online version of the paper at
Acknowledgements We thank B. Sakmann for suggesting the juxtacellular
stimulation approach, J. van der Burg, E. Haasdijk and G. Maas for technical
contributions, P. den Iseger and A. Lee for discussions, and G. Borst, M. Frens,
C. Hansel, L. Herfst, A. Lee, B. Voigtand J. Wolfe for comments on the manuscript.
and Humboldt University Berlin, Erasmus MC, and VIDI (NWO) and HFSP grants
Author Information Reprints and permissions information is available at
www.nature.com/reprints. Correspondence and requests for materials should be
addressed to A.R.H. (firstname.lastname@example.org) or M.B.
NATURE|Vol 451|3 January 2008
Experimental procedures. We used standard surgical and electrophysiological
techniques to prepare animals (n515 Wistar rats, about P35 at the day of
surgery) for chronic, head-fixed recording of the barrel cortex (P3, L5 relative
to bregma)20. Glass pipettes for single-cell stimulation were filled either with
intracellular solution containing (in mM): K-gluconate 135, HEPES 10, Na2-
phosphocreatine 10, KCl 4, MgATP 4, and Na3GTP 0.3 (pH7.2), or Ringer’s
solution containing NaCl 135, KCl 5.4, HEPES 5, CaCl21.8, and MgCl21
(pH7.2). Penetrations were targeted to the E, D and C whisker row representa-
tion of barrel cortex. All experimental procedures were performed according to
Dutch and German guidelines on animal welfare under the supervision of local
Analysis. Cells were included if at least five catch trials and five single-cell
stimulation trials had been presented that satisfied our inclusion criterion. All
single-cell stimulation and catch trials were included for which the animal
responded to both the preceding and the succeeding microstimulation. We also
included all trials where the animal responded only to either the immediately
preceding or immediately succeeding microstimulation. All numbers on single-
cell stimulation experiments refer to these included trials. For our barrel cortex
experiments an average of 30.2 single-cell stimulation trials and 17.7 catch trials
were included per cell. Applying the inclusion criterion led to the exclusion of
many trials with low response rate (on average 14.0 single-cell stimulation trials
and 7.7 catch trials). When all trials were considered, animals still responded
significantly more often in single-cell stimulation trials than in catch trials (data
approximately 480 single-cell stimulation trials (about 16 times more than our
sided binomial test by normal approximation, a50.05) assuming a 10% effect
size (30% hits, 20% false positives).
As we trained animals to report stimulation of the barrel cortex, we tested the
prediction that single-cell stimulation led to responses (hits). Thus, differences
between hit rates and false-positive rates were evaluated by using a one-sided,
paired t-test, and differences between single-cell stimulation experiments and
control experiments by using a one-sided, unpaired t-test. However, all results
presented here were also significant when applying two-sided t-tests or
when using Monte Carlo simulations of the statistical distributions. To test
non-parametrically if the variance of sensory effects was greater for putative
interneurons than excitatory neurons (see Fig. 3e), we ranked all neurons on
effect size and computed the variance of interneuron ranks and compared
this with a Monte Carlo distribution based on random permutations of the
It is our impression that the interneuron inclusion criteria (action potential
width no greater than 0.4ms and/or a response of at least 50 action potentials
to the false classification of potential interneurons as excitatory cells. The differ-
ence in sensory effects between putative interneurons and excitatory cells per-
sisted when more inclusive interneuron inclusion criteria were chosen (data not
shown). Further observations support our classification scheme. First, morpho-
logically identified excitatory neurons were correctly classified as putative excit-
atory cells (Supplementary Fig. 8). Second, maximum discharge patterns of
putative excitatory cells showed significantly more irregular spike intervals than
those of putative interneurons (P50.037, two-sided t-test on the coefficient
of variation), which showed very little or no accommodation (Supplementary
a 30s period at the onset of each experiment before any current injections) were
similar between putative interneurons (mean 2.1Hz) and putative excitatory
Histology. On the last day of experiments with an animal, we included biocytin
Even though combining cell recovery and behavioural measurements was dif-
ficult,we recoveredten neurons orfragments ofneurons.Three ofthese entered
excitatory neurons (see Supplementary Fig. 8). Combining cell recovery and
behavioural measurements is difficult because one is limited to a single session
which complicates neuron identification. We evaluated the tissue damage by
electrode constructs in 8 out of 15 animal brains. In these brains the damage to
the barrel cortex target area was rated as either minimal or weak.
31. Connors, B. W. & Gutnick, M. J. Intrinsic firing patterns of diverse neocortical
neurons. Trends Neurosci. 13, 99–104 (1990).