Letter https://doi.org/10.1038/s41586-018-0527-y
Sensation, movement and learning in the absence of
barrel cortex
­Y. Kate Hong1
, Clay O. Lacefield1
, Chris C. Rodgers1
& Randy M. Bruno1
*
For many of our senses, the role of the cerebral cortex in detecting
stimuli is controversial1–17
. Here we examine the effects of both
acute and chronic inactivation of the primary somatosensory
cortex in mice trained to move their large facial whiskers to detect
an object by touch and respond with a lever to obtain a water
reward. Using transgenic mice, we expressed inhibitory opsins in
excitatory cortical neurons. Transient optogenetic inactivation of
the primary somatosensory cortex, as well as permanent lesions,
initially produced both movement and sensory deficits that
impaired detection behaviour, demonstrating the link between
sensory and motor systems during active sensing. Unexpectedly,
lesioned mice had recovered full behavioural capabilities by the
subsequent session. This rapid recovery was experience-dependent,
and early re-exposure to the task after lesioning facilitated recovery.
Furthermore, ablation of the primary somatosensory cortex
before learning did not affect task acquisition. This combined
optogenetic and lesion approach suggests that manipulations of the
sensory cortex may be only temporarily disruptive to other brain
structures that are themselves capable of coordinating multiple,
arbitrary movements with sensation. Thus, the somatosensory
cortex may be dispensable for active detection of objects in the
environment.
Sensory detection tasks have become a staple for probing cortical
circuitry during behaviour, but the role of the primary sensory cortex
in visual1–3
, auditory4–6
, gustatory7,8
and somatosensory behav-
iours9–17
remains unclear. Whether a brain structure is necessary for
a behaviour is typically assessed by inactivation or ablation. Ablation
experiments may underestimate behavioural deficits, owing to the
long recovery periods used (more than one week), during which compensatory
relearning or rewiring can occur. Transient optogenetic or
pharmacological manipulations often yield stronger deficits and are
currently preferred, being thought to reveal an area’s normal function
before compensation. However, the sudden loss of a silenced area may
disrupt downstream areas that are vital for behaviour, a phenomenon
1
Department of Neuroscience, Mortimer Zuckerman Mind Brain Behavior Institute and Kavli Institute for Brain Science, Columbia University, New York, NY, USA. *e-mail: randybruno@columbia.edu
a
High-speed
camera
Pole
Laser
Lever
Water
Lever
WaterMotor
Pole
Optic fibre
C2
0.3
0.2
0.1
0
Responselatency(s)
j
OFF ON
100
75
50
Percentcorrect
g
OFF ON
***
Falsealarmrate
1.0
0.5
0
i
OFF ON
**
h
OFF ON
Hitrate
1.0
0.5
0
***
stop–Halo
Emx1–Halo
Time (s)
0 1 2
c
Trial
0 1 2
Time (s)
Firingrate(Hz)
0 1 2
Time (s)
d
0
10
20
OFFON
Laser Laser
Lever Lever
Trial
OFFON
30
Firingrate(Hz)
0
10
20
30
e
f
b
Pole Pole
Firingrate(Hz)
0 250
Time from contact (ms)
0
200
400
Firingrate
(%ofbaseline)
600
OFF ON
0
10
20
30
Fig. 1 | Transient inactivation of barrel cortex impairs whiskermediated
detection. a, Head-fixed detection task. A high-speed camera
imaged whiskers. Pole movement (GO or NOGO) and laser (ON or OFF)
were randomized across trials. b, Coronal section of Emx1–GFP mouse
brain. White dotted lines, barrel cortex boundaries; green, GFP; magenta,
NeuN. Scale bar, 1 mm. c, Cortical array recordings during detection
task and photoinactivation. Example rasters (top) and peri-stimulus time
histograms (middle) for a single unit when no response was made (miss
or correct reject) during an example session. d, Same neuron for trials
with lever response (hit or false alarm). The laser turns off when the
mouse responds, ending the trial. Trials sorted by response time, which
varied (arrows in schematic at bottom). e, Effect of laser on neuronal
spiking (n = 62 putative excitatory neurons, 8 sessions, 3 mice; mean ± s.d.
7.02 ± 6.81 and 0.16 ± 0.94 Hz for laser OFF versus ON, respectively).
f, Average spiking activity aligned to first whisker contact of trial during
an example session (n = 10 neurons). Contact times are defined as the
local maxima, rather than onset, of curvature change (Extended Data
Fig. 3; see Methods). Firing rates were normalized to mean rate during
100 ms before contact during laser-OFF trials. Thin lines, individual cells;
thick lines, means of OFF (black) and ON (orange). g–i, Behavioural
performance for laser OFF versus ON trials. g, Per cent correct trials.
h, Hit rate. i, False alarm rate. j, Response latency for hit trials. Emx1–Halo
(n = 10 mice, red), negative control (n = 7 cre-negative, stop–Halo mice,
black). Data shown as mean ± s.e.m. **P < 0.01, ***P < 0.001.
N A t U r e | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
known as diaschisis. Recent studies in the motor system have underscored
how the off-target effects of transient inactivation can lead to
false conclusions18
. To address these disparate outcomes, we compare
transient and chronic manipulations of the barrel cortex subdivision of
the primary somatosensory cortex (S1) during a simple detection task.
Water-restricted mice were trained in the dark to perform a
GO/NOGO sensory detection task with their C2 whisker (Fig. 1a).
Mice self-initiated trials by holding down a lever. On GO trials, a pole
moved within reach of the whisker when protracted. Mice had to
release the lever when they detected the pole to obtain a water reward
(hit). On NOGO trials, the pole moved away from the mouse. Incorrect
responses to NOGO trials (false alarms) were punished with a timeout.
Misses and correct rejects were neither rewarded nor punished.
Conventionally, cortex is optogenetically silenced by activating
inhibitory cells with channelrhodopsin (ChR), but this may inadvertently
stimulate long-range inhibitory connections. We therefore
developed an approach to directly silence excitatory cells by stably
expressing halorhodopsin (Halo) in cortical excitatory neurons. Emx1IRES-cre
mice express Cre recombinase in excitatory cortical neurons
while excluding subcortical structures (Fig. 1b). Halorhodopsin can
be targeted to these neurons by crossing Emx1-IRES-cre with stopHalo
reporter mice (Emx1–Halo). Optogenetic silencing (by shining
a laser onto these neurons during behaviour) was highly efficacious,
blocking 95 ± 4% (mean ± s.e.m.) of spikes in putative excitatory cells
(Fig. 1c–e, Extended Data Fig. 1j, k) including during whisker contacts,
which normally strongly activate barrel cortex (Fig. 1f). Optogenetics
efficiently silenced spontaneous and sensory-evoked activity in neurons
across all cortical layers within a 1-mm radius, encompassing nearly all
of the barrel columns that represent the large facial whiskers (Extended
Data Fig. 1a–i).
Inactivation of barrel cortex significantly reduced overall performance
(P = 3.6 × 10−4
; n = 10 mice; Fig. 1g, Extended Data
Fig. 2a), but performance remained significantly above chance (50%;
P = 1.1 × 10−8
, one-sample t-test). Despite the fact that mice made
fewer responses to both GO and NOGO trials (Fig. 1h, i), response
time was unaffected by the laser (Fig. 1j), demonstrating that the
mice could still manoeuvre the lever. Control mice lacking Cre were
behaviourally unaffected by the laser (n = 7 stop–Halo mice; Fig. 1g–j).
Conventional photoinhibition by activating parvalbumin (PV)positive
inhibitory cells (n = 5 PV–ChR mice) yielded similar results
to Emx1–Halo (Extended Data Fig. 2a). Thus, transient silencing of
barrel cortex impaired detection.
Touch is an active process, during which subjects adjust their movements
as they contact objects in their environment19–21
. Even small
changes in whisking could alter perception19
. Although activation
of barrel cortex can trigger whisker movements16,22
, the effects of
inactivation are less well understood12,23
. We tracked whisking using
high-speed videography (Fig. 2a, b). During NOGO trials, in which
no contacts were possible, barrel cortex inactivation slightly but significantly
decreased protraction velocity, whisker angle and peak
amplitude (Fig. 2c–f, P = 6.5 × 10−3
, 1.4 × 10−3
and 2.5 × 10−2
, for d, e
and f, respectively). Similarly, during GO trials, when the pole was
present inactivation of barrel cortex decreased peak protraction velocity
(Fig. 2g). We found no significant changes in whisking setpoint or frequency.
We also assessed changes in whisker curvature, a proxy for
contact force24
(Extended Data Fig. 3). Small changes in whisker movement
had a large effect on whisker contacts, resulting in a reduction in
force (Fig. 2h) and an increase in the number of trials without contacts
(Fig. 2i). Thus, silencing of the sensory cortex reduced the vigour of
whisker movement.
θ
0 8642
Presponse
a
b
Max | curvature | (mm–1) Presponse
j
0 0.05 0.15
0.5
0
1.0
0.5
0
1.0
No. contacts per trial
k m
0
0.06
0.12
|K|threshold
OFF ON OFF ON OFF ON OFF ON
0
20
40Slope
Slope
0
0.5
1.0
0
3
6
nl o
Contactthresh.
*** **
i
ON
Peak amplitude (°)
g
0 200 400 600 800 1,000
Time from trial start (ms)
c d
ON
Peak velocity (° s–1)
0 2,500 5,000
***
0
2,500
5,000
f
ON
Trials w/o contacts (%)
0 25 50
*
0
25
50
ON
e
0
0.05
0.10
Max | K| (mm–1)
0 0.05 0.1
*
h
0.10
0 1,000 2,000
0
1,000
2,000
Peak velocity (° s–1)
ON
**
80 120
80
120
160
Max angle (°)
ON
**
160
*
30
15
0
0 3015
400 6
NOGO
GO
100
130
100
130
Whiskerangle(°)
OFF
ON
NOGO
GO
Pole
Laser
Lever
Water
OFFOFFOFF
OFFOFFOFF
Fig. 2 | Transient optogenetic inactivation of barrel cortex alters
whisking kinematics and sensory threshold. a, b, High-speed video
frame depicting traced C2 whisker during NOGO trial (a, pole moves
away) and GO trial (b, pole within whisker reach). Whisker position was
measured as its angle (θ) relative to the face. The whisker bends upon
contacting the pole, changing whisker curvature. c, Average whisker
angle for NOGO and GO trials for an example session. A 200-ms window
(blue shaded area), from when the pole was within reach and before the
response, was analysed. Green, average response time. d–f, Whisking
kinematics for each mouse during NOGO trials. d, Peak angular velocity
of whisker protraction; e, maximum whisker angle; f, mean peak whisking
amplitude. g–i, For GO trials: g, peak angular protraction velocity;
h, average maximum change in curvature (ΔΚ); i, per cent of trials
without any contacts. j, Logistic regression of response probability given
maximum change in curvature for an example session. Tick marks indicate
responses (0 for no response; 1 for lever response) on individual trials.
Detection threshold was defined as the value at which response probability
is 0.5. k, Detection threshold for maximum ΔΚ of each mouse. Thick
orange line, means. l, Slope (sensitivity). m, Logistic regression of response
probability against number of contacts per trial for an example session.
n, o, Detection threshold (n) for number of contacts and slope (o).
*P < 0.05, **P < 0.01, ***P < 0.001, n = 10 mice.
N A t U r e | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
We investigated whether behavioural impairment was simply due to
altered whisking or whether there was an accompanying sensory deficit:
whether, for any given stimulus strength, transient inactivation of the
sensory cortex decreased the probability of response. Cortical silencing
significantly increased detection threshold (0.5 response probability) for
curvature (Fig. 2j, k; P = 9.7 × 10−4
) and number of contacts (Fig. 2m, n;
P = 3.3 × 10−3
), but not sensitivity (Fig. 2l, o). We observed similar
motor and sensory deficits in PV–ChR mice (Extended Data Fig. 2b, c).
Thus, transient optogenetic manipulations impair behaviour by both
increasing sensory threshold and decreasing whisker movement.
Increased sensory threshold is distinct from an absolute inability
to detect stimuli. The observed threshold shift could reflect incomplete
inactivation, as a few renegade spikes may suffice for detection25
.
However, residual spiking during optogenetic silencing did not correlate
with behavioural outcome (Extended Data Fig. 1l, m). To ensure
complete inactivation, we removed contralateral barrel cortex by
aspiration (n = 11) (Fig. 3a, b, Extended Data Fig. 4). Consistent with
optogenetic results, behaviour was impaired one day after lesioning
contralateral barrel cortex (Fig. 3c, red), but not in sham-operated controls
(n = 4, black) or when ipsilateral barrel cortex was lesioned (n = 4,
blue). Again, impairment was only partial, and behaviour remained
significantly above chance levels (P < 10−6
, one-sample t-test).
Unexpectedly, by the second session after lesioning, behaviour had
recovered fully to pre-lesion levels (Fig. 3c). Mice recovered whether
lesions encompassed only barrel cortex or additionally included the
secondary somatosensory cortex (Extended Data Fig. 4). There was
no evidence of gradual relearning within sessions; rather, performance
abruptly recovered between the first and second post-lesion sessions
(Fig. 3d), suggesting that recovery was unlikely to result from previously
uninvolved circuits learning the task anew. Furthermore, sub­sequent
lesioning of ipsilateral barrel cortex did not perturb performance,
as these bilaterally lesioned mice performed similarly to sham and
ipsilateral-only lesioned mice (Extended Data Fig. 5). Notably, additional
damage to the dorsolateral striatum prevented behavioural recovery
(Fig. 3c, n = 8, orange; Extended Data Fig. 6), suggesting that the striatum
has an important role in detection behaviour.
Consistent with optogenetic results, whisker movement and contacts
were decreased during the first session after lesioning (Fig. 3e, f, pre
versus1).However,whiskingkinematicsfortherecovered,secondsession
never exceeded pre-lesion levels (Fig. 3e, f; pre versus 2), indicating that
mice did not compensate for impaired sensation with greater contact
force or frequency. Similarly, sensory thresholds pre-lesion and on the
second post-lesion session did not differ significantly in contact force
or number (Fig. 3h, i). Thus, after only a temporary impairment, both
motor and sensory abilities returned to pre-lesion levels along with
behavioural performance.
Recent studies have suggested that homeostasis may spontaneously
restore activity in connected structures within 24–48 h, similar to the
time frame seen here18,26
. To test whether behavioural recovery was
spontaneous or required re-exposure to the task, we gave another
group of mice three days between lesion and retesting (n = 8). This
group showed similar impairment on the first post-lesion session but
also recovered (Fig. 3j), albeit more gradually, indicating that task
re-exposure—rather than simply the passage of time—triggers recovery.
Removing the C2 whisker reduced the performance of both groups
(1 day rest and 3 days rest) to chance, confirming that lesions did not
induce mice to switch from whisker-mediated touch to other sensory
modalities (Fig. 3g, k).
b
h k
a
j
c
Pre 1 2
e
Velocity(°s–1)
GO
f
NOGO
f
d
Max | curvature| (mm × 10–2)
P(response)
1.0
0
0.5
Session from lesion
Percentcorrect
50
75
100
Pre-lesion
1st session
2nd session
2080 4 1612 –2 1–1 2 3 4
3 days rest
i
Curvaturethreshold(mm–1)
No.contactsthreshold
Pre 1 2 Pre 1 2
Sham
Contra
Pre 1 2
Peakamplitude(o)
Nocontacts(%trials)
ΔCurvature(mm–1)
50
75
100
Percentcorrect
***
50
75
100
Percentcorrect
C2 Trim
***
C2 Trim
g
0
1,000
2,000
3,000
10
15
20
25
0.04
0.06
0.08
0.1
0
20
40
60
0.02
0.04
0.06
0.08
*
1.0
1.5
2.0
2.5
Pre 1 2 Pre 1 2
10 20 30
Time within session (min)
50
100 Prelesion 1st session 2nd session
Percentcorrect
75
Sham
Contra S1
10 20 30 10 20 30
Sham
Contra S1
Ipsi S1
Contra S1+ striatum
Percentcorrect
50
75
100 1 day rest
Session from lesion
–2 –1 1 2 3 4
Fig. 3 | Behavioural performance recovers rapidly after barrel cortex
lesions. a, Coronal section of L4-labelled mouse brain (white dashed
lines, barrel cortex boundaries; blue, DAPI; green, eYFP) in unlesioned
hemisphere. b, Lesioned hemisphere shows complete removal of barrel
cortex. c, Behavioural performance before and after S1 lesions (sham
n = 5, ipsilateral n = 4, contralateral n = 11, contralateral with striatal
lesion n = 8). d, Behavioural performance of contralateral S1 lesioned
mice recovers abruptly between first and second sessions after lesioning.
e, f, Whisking kinematics during NOGO trials (e) and GO trials of mice
(f) are altered on first post-lesion session but return to normal by second
post-lesion session (n = 10 contralaterally lesioned mice). g, Post-lesion
trimming confirms that task remained whisker-dependent. Dashed line
indicates chance performance. h, Example logistic regression of response
probability given whisker curvature. Sensory detection threshold for
whisker curvature increases on first session post-lesion (pink) and returns
to pre-lesion levels (black) by second session (red). i, Average detection
thresholds for whisker curvature and number of contacts before lesion and
for second session after lesion (n = 10). j, Mice with three days of rest after
lesion still had impaired performance on the first post-lesion session but
subsequently recovered (n = 8). k, Behavioural performance for three-day
rest group remains whisker-dependent. ***P < 0.001.
N A t U r e | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Thus, three different manipulations of barrel cortex—Emx1–Halo,
PV–ChR and lesions—transiently disrupt the execution of active
detection behaviour. Recent studies have shown that the motor cortex
might not be required for the execution of skilled movements, but is
required for motor learning27
. To determine whether the sensory cortex
is required for learning of the detection task, we lesioned contralateral
S1 of naive mice. Subjects were habituated and trained on lever manoeuvring.
Learning rate was assessed, starting with the introduction to
the pole (Fig. 4a, learning assessment, red). Initially, sessions consisted
of 90% GO trials until mouse weight stabilized (3.6 ± 1.5 sessions),
after which mice still performed at chance (Extended Data Fig. 7a–c).
Lesioned and unlesioned mice learned at similar rates (Fig. 4b, c). Nonlearners
were equally present in both groups (Fig. 4d; 6/17 unlesioned,
5/14 lesioned, P = 1, Fisher’s exact test), and failure to learn was not
correlated with lesion size (Fig. 4e). In fact, among learners, mice with
larger lesions learned faster than those with smaller lesions (Fig. 4e,
linear regression, P = 0.02). Again, performance remained whiskerdependent
(Fig. 4f, g). Thus, barrel cortex is not essential for learning
the detection behaviour.
Task acquisition involves motor (lever press or lift), perceptual
(pole detection) and contingency learning (lever→reward,
contact→lever→reward). Notably, mice acquired the task whether they
were lesioned before handling and lever training (Fig. 4a, open triangle,
n = 4) or just before introduction of the pole (closed triangle, n = 5). Mice
spent similar amounts of time in pre-training whether lesioned before
pre-training or unlesioned (P = 0.13). Thus, barrel cortex appears not to
be required for motor, perceptual and contingency learning of this task.
Our results demonstrate the potential for structures other than
the sensory cortex to direct learned behaviours that require arbitrary
coordination of multiple movements (lever press, whisking and licking)
around a sensory event. This raises concerns about the interpretation
of cortical physiology studies that use detection behaviours, as well as
tasks requiring discrimination of elementary features encoded at the
periphery (body location, retinotopic location, taste and sound frequency),
which subjects may circumvent with detection strategies. It
underscores the need to identify the behavioural conditions for which
sensory cortex is indispensable, which might involve more complex
discrimination, egocentric or allocentric context10,28
, or working
memory29
.
In conclusion, impairment after transient inactivation does not absolutely
indicate necessity. This raises the question of  what the functional
relevance of barrel cortex is to active detection. One possibility
is that the barrel cortex and other structures are redundant for active
detection30
. Multiple subcortical structures receive barrel cortex input
and, via other routes, whisker-related sensory signals. The trigeminal
brainstem complex projects directly to the superior colliculus and cerebellum,
and indirectly to the dorsolateral striatum via the secondary
somatosensory thalamus30–32
. Indeed, damage to striatum prevented
recovery. Further studies are needed to assess the roles of other subcortical
areas.
A second possibility is that manipulation of any cortical area may
temporarily disrupt connected structures that are primarily involved
in the task. In this scenario, the sudden loss of barrel cortex activity,
rather than the sensory information it conveys, can be disruptive. The
incomplete behavioural impairments we observed, as well as the sudden
recovery after lesioning, raise the possibility of a disruptive effect,
rather than redundancy. Subcortical systems are major targets of deep
layer cortical pyramidal cells, which have high baseline firing rates, and
removing their tonic activity may disrupt the responses of corticofugal
targets to sensory inputs. In the birdsong motor system, lesions transiently
disrupt activity in downstream areas and the production of song,
both of which recover overnight18
. In the birdsong study, it was unclear
whether recovery was spontaneous or required some attempts, even
if unsuccessful, at singing. A major advantage of our study is that we
could control the time of re-exposure to the task after lesioning. Early
task-specific experience accelerated recovery, and this may have important
implications for early rehabilitation after stroke or head trauma.
Whether recovery is always experience-dependent or whether sensory
and motor systems differ are intriguing questions for further study.
Online content
Any methods, additional references, Nature Research reporting summaries, source
data, statements of data availability and associated accession codes are available at
https://doi.org/10.1038/s41586-018-0527-y.
Received: 28 September 2017;Accepted: 27 July 2018;
Published online xx xx xxxx.
	1.	 Glickfeld, L. L., Histed, M. H. & Maunsell, J. H. R. Mouse primary visual cortex is
used to detect both orientation and contrast changes. J. Neurosci. 33,
19416–19422 (2013).
	2.	 Petruno, S. K., Clark, R. E. & Reinagel, P. Evidence that primary visual cortex is
required for image, orientation, and motion discrimination by rats. PLoS ONE 8,
e56543 (2013).
	3.	 Lashley, K. S. The mechanism of vision IV. The cerebral areas necessary for
pattern vision in the rat. J. Comp. Neurol. 53, 419–478 (1931).
	4.	 Kato, H. K., Gillet, S. N. & Isaacson, J. S. Flexible sensory representations in
auditory cortex driven by behavioral relevance. Neuron 88, 1027–1039 (2015).
	5.	 Kelly, J. B. & Glazier, S. J. Auditory cortex lesions and discrimination of spatial
location by the rat. Brain Res. 145, 315–321 (1978).
	6.	 Talwar, S. K., Musial, P. G. & Gerstein, G. L. Role of mammalian auditory cortex in
the perception of elementary sound properties. J. Neurophysiol. 85, 2350–2358
(2001).
	7.	 Oliveira-Maia, A. J. et al. The insular cortex controls food preferences
independently of taste receptor signaling. Front. Syst. Neurosci. 6, 5 (2012).
	8.	 Peng, Y. et al. Sweet and bitter taste in the brain of awake behaving animals.
Nature 527, 512–515 (2015).
	9.	 Waiblinger, C., Brugger, D. & Schwarz, C. Vibrotactile discrimination in the rat
whisker system is based on neuronal coding of instantaneous kinematic cues.
Cereb. Cortex 25, 1093–1106 (2015).
	10.	 Hutson, K. A. & Masterton, R. B. The sensory contribution of a single vibrissa’s
cortical barrel. J. Neurophysiol. 56, 1196–1223 (1986).
403020
Sessions to criterion
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Lesioned (n = 14)
Unlesioned (n = 17)
e
a
******
Cum.fractionofmice
50
100
Percentcorrect
75
C2 Trim C2 Trim
No. of sessions to criterion
0 10
f
Lesiondiameter(mm)
0
0.5
1.0
1.5
2.0
2.5 Fractionofnon-learners
+–
50
100Percentcorrect
75
0
0.5
1.0
b
Handling Lever
training
~4 ~9
No. of
sessions
Pretraining Learning assessment
d
g
c
0
10
20
30
40
50
Lesion +–
Sessionstocriterion
< 48
90% GO
50
4030200 10 50 NL
~4
60% GO
Fig. 4 | Barrel cortex is not required for learning whisker-dependent
sensory detection task. a, Timeline for learning assay. b, Cumulative
histogram of number of sessions to reach learning criterion (>74% correct
for two consecutive sessions). The speed of learning was similar between
unlesioned (n = 17) and lesioned (n = 14) mice. c, Among mice that
learned the task, the number of sessions to criteria did not differ between
unlesioned (n = 11) and lesioned (n = 9) mice. d, The fraction of mice that
did not learn within the assessment period was similar between groups
(unlesioned n = 6, lesioned n = 5 mice). e, Inability to learn was not due to
larger lesion size. NL, non-learners. f, g, Unlesioned (f) and lesioned mice
(g) that did learn could no longer perform the task when the C2 whisker
was trimmed. ***P < 0.001.
N A t U r e | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
	11.	 Morita, T., Kang, H., Wolfe, J., Jadhav, S. P. & Feldman, D. E. Psychometric curve
and behavioral strategies for whisker-based texture discrimination in rats. PLoS
ONE 6, e20437 (2011).
	12.	 Guo, Z. V. et al. Flow of cortical activity underlying a tactile decision in mice.
Neuron 81, 179–194 (2014).
	13.	 O’Connor, D. H., Peron, S. P., Huber, D. & Svoboda, K. Neural activity in barrel
cortex underlying vibrissa-based object localization in mice. Neuron 67,
1048–1061 (2010).
	14.	 Kwon, S. E., Yang, H., Minamisawa, G. & O’Connor, D. H. Sensory and decisionrelated
activity propagate in a cortical feedback loop during touch perception.
Nat. Neurosci. 19, 1243–1249 (2016).
	15.	 Miyashita, T. & Feldman, D. E. Behavioral detection of passive whisker stimuli
requires somatosensory cortex. Cereb. Cortex 23, 1655–1662 (2013).
	16.	 Sachidhanandam, S., Sreenivasan, V., Kyriakatos, A., Kremer, Y. & Petersen, C. C.
H. Membrane potential correlates of sensory perception in mouse barrel cortex.
Nat. Neurosci. 16, 1671–1677 (2013).
	17.	 Stüttgen, M. C. & Schwarz, C. Barrel cortex: what is it good for? Neuroscience
368, 3–16 (2018).
	18.	 Otchy, T. M. et al. Acute off-target effects of neural circuit manipulations. Nature
528, 358–363 (2015).
	19.	 Deutsch, D., Pietr, M., Knutsen, P. M., Ahissar, E. & Schneidman, E. Fast feedback
in active sensing: touch-induced changes to whisker-object interaction. PLoS
ONE 7, e44272 (2012).
	20.	 Grant, R. A., Mitchinson, B., Fox, C. W. & Prescott, T. J. Active touch sensing in the
rat: anticipatory and regulatory control of whisker movements during surface
exploration. J. Neurophysiol. 101, 862–874 (2009).
	21.	 Mitchinson, B. et al. Active vibrissal sensing in rodents and marsupials.
Phil. Trans. R. Soc. Lond. B 366, 3037–3048 (2011).
	22.	 Matyas, F. et al. Motor control by sensory cortex. Science 330, 1240–1243
(2010).
	23.	 Harvey, M. A., Sachdev, R. N. & Zeigler, H. P. Cortical barrel field ablation and
unconditioned whisking kinematics. Somatosens. Mot. Res. 18, 223–227
(2001).
	24.	 Pammer, L. et al. The mechanical variables underlying object localization along
the axis of the whisker. J. Neurosci. 33, 6726–6741 (2013).
	25.	 Stüttgen, M. C. & Schwarz, C. Psychophysical and neurometric detection
performance under stimulus uncertainty. Nat. Neurosci. 11, 1091–1099 (2008).
	26.	 Keck, T. et al. Synaptic scaling and homeostatic plasticity in the mouse visual
cortex in vivo. Neuron 80, 327–334 (2013).
	27.	 Kawai, R. et al. Motor cortex is required for learning but not for executing a
motor skill. Neuron 86, 800–812 (2015).
	28.	 Brecht, M. The body model theory of somatosensory cortex. Neuron 94,
985–992 (2017).
	29.	 Stüttgen, M. C., Schwarz, C. & Jäkel, F. Mapping spikes to sensations. Front.
Neurosci. 5, 125 (2011).
	30.	 Cohen, J. D. & Castro-Alamancos, M. A. Detection of low salience whisker stimuli
requires synergy of tectal and thalamic sensory relays. J. Neurosci. 30,
2245–2256 (2010).
	31.	 Huerta, M. F., Frankfurter, A. & Harting, J. K. Studies of the principal sensory and
spinal trigeminal nuclei of the rat: projections to the superior colliculus, inferior
olive, and cerebellum. J. Comp. Neurol. 220, 147–167 (1983).
	32.	 Smith, J. B., Mowery, T. M. & Alloway, K. D. Thalamic POm projections to the
dorsolateral striatum of rats: potential pathway for mediating stimulus-response
associations for sensorimotor habits. J. Neurophysiol. 108, 160–174 (2012).
Acknowledgements We thank B. C. Pil and A. Kase for assistance with mouse
training, H. Zeng and M. Kheirbek for Ai39 mice, A. Losonczy for PV–Cre mice,
J. Huang for Rosa–H2B–GFP mice, N. Clack for suggestions on video analysis
of whisker motion, B. Grewe for advice on cortical aspiration, and J. C. Tapia,
M. Goldberg, N. Sawtell, T. Jessell, A. Kinnischtzke, D. Kato and G. Pierce for
comments on the manuscript. Funding was provided by NIH R01 NS094659,
R01 NS069679, the Klingenstein Fund, the Rita Allen Foundation, the Dana
Foundation and the Ludwig Schaefer Scholars Program (R.M.B.); NIH F32
NS084768 (Y.K.H.), T32 MH015174 (Y.K.H., C.O.L.) and F32 NS096819
(C.C.R.).
Reviewer information Nature thanks C. Schwarz and the other anonymous
reviewer(s) for their contribution to the peer review of this work.
Author contributions Y.K.H. and R.M.B. conceived the experiments, analysed
data and wrote the manuscript. Y.K.H. performed the experiments. C.O.L. and
Y.K.H. developed the behavioural assay. C.C.R. assembled videography setup,
C.C.R. and Y.K.H. performed array recordings, and C.C.R. analysed array data.
Competing interests The authors declare no competing interests.
Additional information
Extended data is available for this paper at https://doi.org/10.1038/s41586-
018-0527-y.
Supplementary information is available for this paper at https://doi.org/
10.1038/s41586-018-0527-y.
Reprints and permissions information is available at http://www.nature.com/
reprints.
Correspondence and requests for materials should be addressed to R.M.B.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
N A t U r e | www.nature.com/nature
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Methods
Transgenic mice. All mouse procedures complied with the NIH Guide for the Care
and Use of Laboratory Animals and were approved by the Institutional Animal
Care and Use Committee at Columbia University. Emx1–Halo mice (n = 10) were
generated by crossing Emx1-IRES-Cre33
knock-in mice (Jackson Laboratories,
stock #005628) to Rosa-lox-stop-lox (RSL)-eNpHR3.0/eYFP mice (stop-Halo
Ai39, JAX, stock# 006364), which express halorhodopsin after excision of a stop
cassette by Cre recombinase. Cre expression was assessed by crossing Emx1–Cre
mice to RSL-H2B–GFP mice (provided by J. Huang). Negative control mice were
Cre-negative and could not express the halorhodopsin transgene (n = 7 stopHalo
mice). PV–ChR mice (n = 5) were generated by crossing parvalbumin-Cre
mice34
to RSL-channelrhodopsin2/eYFP mice (Ai32, Jackson Laboratories, stock#
024109). For visualizing barrels in S1, Nr5a1–Cre mice (JAX, stock# 006364) were
crossed to RSL-Halo–eYFP (Nr5a1–eYFP). All mouse lines were maintained on a
C57BL/6 background. Optogenetic experiments used mice that were heterozygous
for the desired transgene. The experimenters were not blind to genotype during
testing and analysis.
Intrinsic signal optical imaging to locate C2 barrel. Mice were anaesthetized
with isoflurane, and the skull over the left barrel cortex (centred ~1.5 mm posterior
to bregma and 3.5 mm lateral to the midline) was thinned and sealed
with Vetbond (3M) over a 4–5-mm area, or a glass window (3-mm coverslip,
Warner Instruments) was implanted. Images were acquired with a CCD camera
(Q-Imaging, Retiga 2000R) mounted on a stereomicroscope and software
custom-written in LabVIEW. The vasculature on the brain surface was imaged
with 510/40 band-pass filtered illumination (Chroma, D510/40), and functional
imaging done with illumination with a 590-nm long-pass filter (Thorlabs, OG590).
The C2 whisker was stimulated with 8 pulses of 4 directions at 5 Hz with a multi-directional
piezo stimulator. The location of the maximum reflectance change
was mapped relative to the surface vasculature. Two or three surrounding whiskers
were also imaged to confirm proper identification of the C2 barrel location.
Behavioural setup. The behavioural setup was controlled by a microcontroller
(Arduino), and data collected using custom-written routines. Subjects selfinitiated
trials by holding down a lever with their left forepaw for at least 100 ms.
A pole (~2.15-mm-diameter wooden applicator stick) started from a position
3–4 cm below the mouse. After trial initiation, a stepper motor (Pololu Robotics
and Electronics) rotated the pole to just in front of the whiskers (GO trials) or
away from the whisker field entirely (NOGO trials). GO and NOGO trials were
randomized. Rotation in either direction ensured that the sound and vibration
generated by the motor was similar between trial types. The sound of the motor
also served as a trial onset and offset cue. For GO trials, the pole was positioned
9–11 mm laterally and roughly aligned to the tip of the nose in the anterior–
posterior dimension, such that mice were required to actively whisk forward to
make contact. If the lever was lifted within the response window of 1.2 s during
GO trials (hit), the response was rewarded with a drop of water and a reward tone.
False alarm responses during NOGO trials were punished with a timeout period
of 3–8 s accompanied by white noise sound, during which the mouse could not
initiate a new trial. There was no reward for a correct reject and no punishment for
a miss. The response latency was defined as the time from when the pole was first
within reach of the mouse’s whisker (typically 480 ms from trial onset, identified
from high-speed video for each session analysed) to the mouse’s lever lift response.
Mouse training. Test of task performance. Adult mice (P116 ± 60 days, mean and
s.d., 34% male) were implanted with a custom-designed 22-gauge stainless steel
laser-cut (Laser Alliance) headplate with dental acrylic. After ~1 week of recovery,
subjects in optogenetics experiments were water-restricted and then trained in
stages. Progression through each stage depended on the individual mouse’s weight
stabilization (indication of health). (1) Freely moving mice were habituated to the
behavioural apparatus, where water reward was given for holding down a lever
(2–3 days). (2) Mice were head-fixed and continued lever training: water was
awarded for holding down lever for 500 ms to initiate trial, followed by releasing
the lever for >100 ms (2–4 days). (3) Mice were trained in a dark chamber where
no visual cues could be used. Mice were trained with 90% GO trials with a pole,
and received water for responding by lifting a lever on GO trials, or a 3–5 s timeout
if response was on a NOGO trial (2–8 days). We found that adding this gradual
training stage with 90% GO trials facilitated learning. (4) Mice were trained at
60% GO trials until learning criterion (defined as >74% correct performance for
2 consecutive days) was reached.
Once subjects learned the detection task with all whiskers intact, the location of
the C2 barrel was functionally mapped using intrinsic signal optical imaging. All
whiskers except C2 on the mouse’s right side were trimmed. Mice were retrained
with a single C2 whisker until >74% performance was reached with 50% GO trials.
In most cases, behavioural performance dropped after the initial whisker trimming
but recovered over 1–7 days. Trimming was maintained twice a week. Lesions or
sham operations were made after the performance of the mice stabilized above
the performance criterion. Experimenters were not blind to whether mice were
lesioned. Any animals that could still perform well (>60% correct) in the absence
of all whiskers were excluded from analysis.
Test of task learning. Experimenters were blind to whether mice were lesioned
or sham. For learning experiments, mice were trained as above but with more
standardized training periods. Mice recovered from surgery and were habituated
to handling by the trainer for 1–2 days. Mice were water-restricted and habituated
through a series of pre-training stages. Progression through each step depended on
the individual mouse’s weight stabilization (indication of health), or whether rest
days (weekends) interrupted training, rather than an explicit behavioural criterion.
(1) Freely moving mice were habituated in the light to the behavioural apparatus,
where water reward was given for holding down a lever (1–2 days). (2) Mice were
head-fixed and continued to hold down the lever to drink (4 ± 2 days, mean and
s.d.). (3) Mice were rewarded for holding down the lever and subsequently lifting
for >100 ms (4 ± 2 days). All whiskers excluding C2 were trimmed twice a week
for the remainder of the learning experiment. We quantified learning time for
the following stages: mice were first introduced to the pole with 90% GO trials
(~4 days, Extended Data Fig. 7). This stage and successive ones were performed in
a light-tight box where no visual cues could be used. Mice were given <48 sessions
to reach learning criterion of 74% correct for 2 consecutive sessions. To factor
in the different number of GO and NOGO trials, we calculated per cent correct
performance as 100 × (Nhit/NGO + NCR/Nnogo)/2 in which the N variables are the
numbers of hit, go, correct reject and NOGO trials, respectively.
A total of 29 mice were lesioned before learning. Of these, several mice were
excluded from analysis, including 6 mice that performed >60% correct after full
whisker trimming; 3 mice that were found to have lesions that did not include the
C2 barrel; and 6 mice that had lesions that extended below the cortex, including
white matter and striatum.
Quantification of whisker movements and contacts. Whisking was monitored
with a high-speed camera (Photonfocus AG, MV1-D1312-100-G2) at 250 fps and
640 × 480 pixels/frame under infrared illumination. Whiskers were automatically
traced offline and whisker position (angle) and curvature were obtained using
Whisk35,36
. Trials in which tracing failed >10% of frames were omitted from analysis.
For each session, mice were given a 5-min warm-up before analysis began,
except for quantification of within-session performance (Fig. 3d) for which the
entire session was included for analysis. Whisking analysis was restricted to a
200-ms window starting from the first frame in which whisker contact with the
pole was possible. This window was chosen to best isolate the whisking associated
with the sampling of the object and before the response, after which, whisking
tended to increase in association with licking for water reward (Fig. 2g). For
whisking amplitude and phase, the azimuthal whisking angle was band-pass
filtered (four-pole Butterworth, 4–50 Hz) followed by a Hilbert transform37,38
.
Instantaneous frequency was calculated from the phase. The setpoint was measured
as the midpoint between the whisking envelope defined by the maximum and
minimum whisker angles for each whisk cycle.
Whisker contacts were defined using whisker angle and curvature parameters
for each session. We first determined the range of angle-curvature values during
free whisking in air (NOGO trials). The baseline curvature for each trial (mean
change in curvature during 200-ms period before each trial onset) was subtracted
to obtain the change in curvature (ΔΚ). Linear regression of the whisker angle
and change in curvature was used to find the line of best fit (Extended Data Fig. 3).
Upper and lower contact thresholds were set by finding the offset of the lines that
encompassed the angle-curvature parameter space for all NOGO trials (1–5 s.d.).
Putative contacts were defined as points at which local maxima or minima of ΔΚ
were above or below the defined contact thresholds.
Optogenetic modulation of cortical activity. For optogenetic experiments during
detection behaviour, the laser was on for 33–50% of trials, which were randomly
interleaved with laser-off trials. For laser-on trials, an optical shutter opened at
the onset of the trial, before movement of the pole. The pole moved within reach
of the whisker field 200–400 ms after the onset of the laser, ensuring photoinactivation
before contacts were possible. The laser remained on until after the mouse
responded, or for the duration of the trial if no response was made, for a maximum
of 1.5 s. Spiking activity during photoinhibition could be efficiently silenced for at
least 2 s; all trials were 1.2–1.5 s in duration with a 1-s inter-trial-interval (Extended
Data Fig. 1). Exposure to laser was limited to minimize photodamage to tissue.
With the protocol described, no physiological or physical damage was detectable.
A 593- or 594-nm laser (OEM or Coherent) was used with Emx1–Halo mice. For
PV–ChR mice, an optical chopper (Thorlabs, MC2000B) modulated the 4 mW
output of a 473-nm laser (OEM) to produce pulses at 40 Hz. Lasers were coupled
to a 200-μm diameter, 0.39 NA optic fibre (Thorlabs) via a fibreport, and the
diamond-knife cut fibre tip was placed above the optical window and positioned
over C2 using the vasculature-referenced intrinsic signal map.
Electrophysiology. For recordings without behaviour, Emx1–Halo mice (n = 4)
were habituated to head-fixation. Juxtasomal recordings were made using pipettes
filled with artificial cerebral spinal fluid and an Axoclamp 900A amplifier. Airpuff
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
stimuli for each test condition were delivered by opening an air valve for 50 ms
during trials ranging from 0.5 to 2 s, and an inter-trial interval of 1.5–3 s for 30–50
trials. Laser-on trials were randomly interleaved for 50% of the trials. Glass pipettes
were inserted perpendicular to the cortical surface (~30° from vertical), and the
optic fibre was positioned vertically near the pipette entry point, above a thinned
skull. To test the effect of photoinhibition at various distances, the optic fibre was
positioned 0, 0.5, 1, 1.5 and 2 mm from the original recording site along a thinned
and transparent skull (n = 21, 11, 10, 12 and 16, respectively). Regular spiking
(RS, putative excitatory) versus fast-spiking (FS, putative inhibitory) cells were
categorized based on their spike waveforms as previously described39
. Cortical
depth was defined as the microdrive depth relative to the pial surface.
For recordings during the behavioural task, a linear silicon array (Cambridge
NeuroTech H3), consisting of 64 sites spanning 1,275 μm, was used. Each site was
11 × 15 μm and coated with PEDOT to obtain an impedance of 50–150 kΩ. Signals
were band-pass filtered 1–7,500 Hz and sampled at 30 kHz (OpenEphys). Between
sessions, the array was withdrawn and the craniotomy sealed with silicone. Spikes
were clustered and inspected using Kilosort40
and Phy41
.
Cortical lesions. Mice were deeply anaesthetized under isoflurane. A 1–4-mm
craniotomy was made and the underlying cortical tissue was aspirated with a sterile
blunt-tipped syringe needle connected to a vacuum. Lesions were made by aspirating
all cortical layers to encompass, at a minimum, the C2 barrel and the immediately
adjacentbarrels,andatamaximum,themajorityofS1representingthelargewhiskers
(macrovibrissae) and secondary somatosensory cortex (Extended Data Figs. 4, 5).
Sham-operated mice were anaesthetized under the same conditions, and the skull
was thinned with a dental drill. Lesioned and sham-operated mice were allowed to
recover for 1 or 3 days after surgery before testing. After behavioural testing was complete,
mice were perfused and brains extracted for histological analysis. Brains were
sectioned tangentially or coronally (100 μm thick) with a vibratome. Lesion diameter
was quantified in ImageJ by outlining the lesioned area for each section, quantifying
the mean Feret diameter and averaging across all sections. Volume was measured
by summing each section’s lesion area multiplied by 100 (the section thickness).
In some cases, lesions extended beyond S1 and into subcortical tissue including
the striatum. To objectively score striatal damage, the extent of cortical and subcortical
damage was scored by five raters experienced at looking at coronal sections
of mouse brains and blind to the behavioural data.
Data analysis. Analyses were done with custom-written scripts in MATLAB. For
all figures, statistical significance is denoted as *P < 0.05, **P < 0.01, ***P < 0.001.
Differences were not significant (P > 0.05) unless otherwise indicated. Nonnormally
distributed data (D’Agostino–Pearson test) were reported using medians
and interquartile ranges, and normally distributed data with means and s.e.m.
Two-sided paired t-tests were used unless otherwise indicated. Based on the mean
and s.d. of normal performance levels of trained mice, power analysis indicated
that detecting a drop in performance to chance levels with a significance criterion
of 0.05 required a minimum sample size of three, which we exceeded in all cases.
Reporting summary. Further information on experimental design is available in
the Nature Research Reporting Summary linked to this paper.
Code availability. All computer codes are available from the corresponding author
upon reasonable request.
Data availability
All data are available from the corresponding author upon reasonable request.
	33.	 Gorski, J. A. et al. Cortical excitatory neurons and glia, but not GABAergic
neurons, are produced in the Emx1-expressing lineage. J. Neurosci. 22,
6309–6314 (2002).
	34.	 Scholl, B., Pattadkal, J. J., Dilly, G. A., Priebe, N. J. & Zemelman, B. V. Local
integration accounts for weak selectivity of mouse neocortical parvalbumin
interneurons. Neuron 87, 424–436 (2015).
	35.	 O’Connor, D. H. et al. Vibrissa-based object localization in head-fixed mice.
J. Neurosci. 30, 1947–1967 (2010).
	36.	 Clack, N. G. et al. Automated tracking of whiskers in videos of head fixed
rodents. PLoS Comput. Biol. 8, e1002591 (2012).
	37.	 Hill, D. N., Curtis, J. C., Moore, J. D. & Kleinfeld, D. Primary motor cortex reports
efferent control of vibrissa motion on multiple timescales. Neuron 72, 344–356
(2011).
	38.	 Kleinfeld, D. & Deschênes, M. Neuronal basis for object location in the vibrissa
scanning sensorimotor system. Neuron 72, 455–468 (2011).
	39.	 Bruno, R. M. & Simons, D. J. Feedforward mechanisms of excitatory and
inhibitory cortical receptive fields. J. Neurosci. 22, 10966–10975 (2002).
	40.	 Pachitariu, M., Steinmetz, N., Kadir, S., Carandini, M. & Harris, K. D. Kilosort:
realtime spike-sorting for extracellular electrophysiology with hundreds of
channels. Preprint at https://www.biorxiv.org/content/early/2016/06/30/
061481 (2016).
	41.	 Rossant, C. et al. Spike sorting for large, dense electrode arrays. Nat. Neurosci.
19, 634–641 (2016).
	42.	 Paxinos, G. & Franklin, K. B. J. The Mouse Brain in Stereotaxic Coordinates.
2nd ed. (Academic, New York, 2001).
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Extended Data Fig. 1 | See next page for caption.
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
Extended Data Fig. 1 | Optogenetic photoinhibition of cortical neurons
in Emx1–Halo mice is highly efficacious. a, Juxtasomal recordings in
awake, head-fixed mice were made below an optic fibre placed above a
thinned, transparent skull. b, Raster plot for example neuron for randomly
interleaved laser off (top) or on (bottom) trials, with air puff schematized
below. c, Population peri-stimulus time histograms of 35 cells with
regular-spiking (RS) waveforms (cortical depth: 280–1,120 µm, n = 4
mice) for laser-off and -on trials. Both spontaneous and whisker stimulusevoked
spikes are silenced. Shaded area: laser on. d, e, Efficiency of RS
cell inactivation as a function of laser power (d) and cortical depth (e),
where per cent inactivation is relative to a cell’s spike rate during laseroff
trials. f, Lateral extent of inactivation. Illumination of 20–40 mW
reliably inactivated an area within a 1-mm radius. g, Photoinhibition
at 40 mW fully blocked spontaneous and sensory-evoked spikes (100%
inactivation relative to laser-off trials) in 83% of RS cells and >96% of
spikes in 94% of RS cells (same cells as in c); P = 3.1 × 10−11
, Wilcoxon
rank-sum test. h, Fast-spiking (FS) neurons (n = 8 cells) were similarly
silenced; P = 3.1 × 10−8
, Wilcoxon rank-sum test. i, Estimated area of
photoinhibition with 40-mW relative to barrel cortex (1-mm radius
around C2 barrel, red circle) depicted with a tangential section through
barrel cortex of an Nr5a1–eYFP mouse with layer 4 labelled to visualize
barrels. j, Emx1–Halo-mediated cortical inactivation was also assessed
during detection behaviour with array recordings (n = 8 session, 3 mice).
k, Data from Fig. 1e replotted on a logarithmic scale to show low spike
rates during laser-on trials. l, Behavioural performance during laser-off
and -on trials did not correlate with spiking activity for each trial type
(four sessions from three mice with per cent correct for laser-off trials:
82, 87, 80, and 78%). m, Laser-on data in l, replotted with larger scale to
visualize data during laser-on trials. Data shown as median ± interquartile
range (d–h); mean ± s.e.m. (l, m).
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Extended Data Fig. 2 | Optogenetic manipulations of barrel cortex
using PV–ChR and Emx1–Halo mice result in similar behavioural,
motor, and sensory deficits. a, Photoinhibition of excitatory neurons in
Emx1–Halo mice (orange) and photoactivation of inhibitory neurons in
PV–ChR mice (blue) yielded similar behavioural deficits. Negative control
mice (Cre-negative, stop-Halo mice; black) were unaffected by 593-nm
laser illumination. P values for Emx1–Halo: hit, 1.6 × 10−5
; false alarm
(FA), 1.7 × 10−3
; per cent correct, 3.6 × 10−4
; d-prime, 1.8 × 10−3
. For
PV–ChR: hit, 3.5 × 10−3
; per cent correct, 3.9 × 10−3
; d-prime, 0.0498.
b, Optogenetic inactivation of S1 with either Emx1–Halo or PV–ChR mice
decreases whisking kinematics. P values for Emx1–Halo: NOGO peak
velocity, 6.6 × 10−3
; max angle, 1.4 × 10−3
; peak amplitude, 2.6 × 10−2
;
GO peak velocity, 4.3 × 10−4
; max change in curvature (ΔΚ), 1.9 × 10−2
;
per cent trials with no contacts, 1.2 × 10−2
. PV–ChR: NOGO peak velocity,
8.1 × 10−3
; max angle, 8.7 × 10−2
; peak amplitude, 8.4 × 10−2
; GO peak
velocity, 5.7 × 10−3
; ΔΚ, 8.7 × 10−3
; per cent trials with no contacts,
7.5 × 10−3
. c, Sensory thresholds increase with optogenetic inactivation.
P values for Emx1–Halo: curvature threshold, 9.7 × 10−4
; contact
threshold, 3.4 × 10−3
. PV–ChR: curvature threshold, 1.1 × 10−1
; contact
threshold, 1.2 × 10−2
. Data for negative control and Emx1–Halo mice are
the same as in Figs. 2, 3 but repeated here for comparison with PV–ChR
mice.
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
Extended Data Fig. 3 | Defining contacts based on whisker angle and
change in curvature. a, Example curvature versus whisker position for a
single session. Each circle represents the paired values for curvature and
whisker angle for each frame during the session. Values for NOGO trials
define whisker parameters during free whisking in air, when no contacts
can be made (black); GO trials are shown in red. Linear regression was
used to define the line of best fit (blue, solid line) for NOGO parameters,
and upper and lower contact thresholds were set by finding the offsets that
encompassed the no-contact parameter space (1–5 s.d. from the line of
best fit, blue dashed lines). b, Putative contacts were defined as points at
which the local maxima or minima of the change in curvature were above
(forward contact with whisker) or below (reverse contact with whisker)
the defined thresholds (tick marks). Whisking analysis was restricted to
the 200-ms time window (yellow shaded area) during sampling, before the
average response time (green).
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Extended Data Fig. 4 | See next page for caption.
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
Extended Data Fig. 4 | Behavioural performance after lesions did
not correlate with lesion size. a, Mice performed the task with a single
C2 whisker. b, The location of the C2 barrel in a coronal section. The
C2 barrel was functionally mapped with intrinsic imaging and Alexaconjugated
cholera toxin subunit B (CTB, red) was injected into the centre
of the C2 barrel. Blue, DAPI. Mappings in b and c are used to inform the
locations of lesions made relative to the C2 barrel column. c, Equivalent
location in section from an Nr5a1–eYFP mouse with barrels fluorescently
labelled in L4 (white) overlaid on bright-field image to show extent of
barrel cortex relative to section (black lines). C2 was located about
1.2–1.5 mm posterior to bregma, varying slightly between mice. Lesions
were centred around C2. d, Size and locations of contralateral barrel cortex
lesions for the 11 mice with 1-day rest shown in Fig. 3 (arranged from
largest to smallest by lesion volume). For each mouse, three locations along
the anterior–posterior axis are shown overlaid on atlas images, reproduced
with permission from ref. 42
. In a few mice (for example, mice 1, 3 and 8),
lesions extended into the secondary somatosensory area (S2). Numbers
along anterior–posterior axis indicate approximate location relative to
bregma. e, Lesion size did not correlate with the degree of impairment
on the first (grey) or second post-lesion session when behavioural
performance had recovered (red). Performance was normalized to the prelesion
performance for each mouse. f, Lesion sizes were similar between
groups with 1 or 3 days of rest after lesioning (P = 0.91).
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Extended Data Fig. 5 | Ipsilateral S1 does not compensate for loss of
contralateral S1. a, Behavioural performance of mice recovered rapidly
after contralateral S1 lesions, as shown in Fig. 3c (red). Subsequent
ipsilateral lesion (n = 6) had effects similar to sham and ipsilateral-only
manipulations shown in Fig. 3c, indicating that ipsilateral S1 was not
compensating for loss of contralateral S1 activity. b, C2 whisker-trim
control. Performance of bilaterally lesioned mice dropped to chance when
the C2 whisker was removed (P = 2.7 × 10−3
). c, Sizes of bilateral lesions.
© 2018 Springer Nature Limited. All rights reserved.
Letter RESEARCH
Extended Data Fig. 6 | Example histology from lesioned mice depicting
lesion of contralateral S1 only or additional damage to striatum.
a, Unlesioned example. b, c, Examples of S1-only lesions from two
mice. d–f, Three examples of damage to striatum in addition to S1.
Even minimal damage (arrows) to the dorsolateral striatum resulted in
permanent behaviour deficits. Scale bar, 1 mm.
© 2018 Springer Nature Limited. All rights reserved.
LetterRESEARCH
Extended Data Fig. 7 | Learning curves for unlesioned and lesioned
mice in learning experiment. a, Individual learning curves for unlesioned
mice (n = 11, blue lines) and mice with lesions of contralateral barrel
cortex (n = 9, orange lines) that learned the detection task to criterion
(74% correct performance for two consecutive sessions). Mice were first
introduced to the pole with 90% GO trials (red lines). This intermediate
step ensured that mice maintained a stable weight before moving on to the
last step of training. Mice were moved onto 60% GO trials for the rest of
the learning assessment. Mice were given 48 sessions to learn the task.
b, Unlesioned and lesioned groups spent similar times on 90% GO sessions
(3.8 ± 1.8 versus 3.5 ± 1.1 sessions for unlesioned and lesioned mice,
respectively; P = 0.64). c, By the end of the 90% GO sessions, performance
was still at chance levels (P = 0.34 and P = 0.37 for unlesioned and lesioned
mice, respectively; one-sample t-test).
© 2018 Springer Nature Limited. All rights reserved.
1
natureresearch|reportingsummaryApril2018
Corresponding author(s): Randy M. Bruno
Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.
Statistical parameters
When statistical analyses are reported, confirm that the following items are present in the relevant location (e.g. figure legend, table legend, main
text, or Methods section).
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
An indication of whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistics including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND
variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Clearly defined error bars
State explicitly what error bars represent (e.g. SD, SE, CI)
Our web collection on statistics for biologists may be useful.
Software and code
Policy information about availability of computer code
Data collection For array recording, open-source software (OpenEphys) was used. The behavior was implemented using custom code, which is available
upon request.
Data analysis Open-source software was used for whisker measurements (Clack et al 2012) and spike sorting and clustering (Kilosort and Phy). All other
analyses were performed with custom MATLAB programs, which are available upon request.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers
upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability
All computer code and data are available from the corresponding author upon reasonable request.
2
natureresearch|reportingsummaryApril2018
Field-specific reporting
Please select the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/authors/policies/ReportingSummary-flat.pdf
Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample size (1) Inactivation of barrel cortex during behavior -- optogenetic inactivation and lesions
Previous studies reported transient inactivation decreased performance to chance levels (50% performance, e. g., O'Connor et al., 2010, Guo
et al., 2014, Sachidhanandam et al., 2013). Based on the mean and standard deviation of normal performance levels of trained animals (81.4%
+/- 7.4%), power analysis yielded a minimum n of 3 to detect a change in performance to chance levels with a significance value of P = 0.05.
In this study, n = 10 Emx-eNphR, n = 7 control animals, and n = 5 PV-ChR mice were included in the study. For lesion studies, n = 8 and n=9
mice were tested for each group (1 and 3-day rest groups).
(2) Learning with lesions
Based on the average learning speed of previously trained animals (25 +/- 8.6 sessions), power analysis indicated a sample size of 5 would be
needed to detect a 2-fold increase in learning speed. In this study, we included 11 unlesioned, and 9 lesioned animals.
Data exclusions Based on preestablished criteria, several conditions led to exclusion of data: (1) animals that were able to perform the detection task (>60%
correct performance) after all whiskers were trimmed were deemed to be using other sensory modalities to perform the task and excluded
from further analysis. (2) For the learning experiments, if animals lost their C2 whisker during training, subjects were excluded from further
analysis. For cortical lesions included in learning experiments, animals that had inadvertent lesions beyond cortex, extending below the white
matter tract were excluded from analysis.
Replication All attempts at replication were successful.
Randomization For learning experiments, cohorts of 10-15 mice from multiple litters were trained at a time. Animals were chosen at random, such that
roughly half of the animals from the same litter (and home cage) received cortical lesions, while the other half received sham-operations. The
first cohort was trained in 2016 by one trainer, and the experiment was repeated in 2017 by a second trainer.
Blinding For learning experiments, the human trainers were blind to which animals received lesions. For analysis of the extent of lesion (cortex only vs.
cortex plus subcortical areas (usually the striatum), histology images from all animals were scored blindly by 5 "experts" who had 3 or more
years of mouse histology experience. Subjects were blind to the animal identity and outcome of the experiments (i.e., recovered behaviorally
or whether they learned the task, if for learning experiments).
Reporting for specific materials, systems and methods
Materials & experimental systems
n/a Involved in the study
Unique biological materials
Antibodies
Eukaryotic cell lines
Palaeontology
Animals and other organisms
Human research participants
Methods
n/a Involved in the study
ChIP-seq
Flow cytometry
MRI-based neuroimaging
Antibodies
Antibodies used Mouse anti-NeuN Antibody, clone A60 (Millpore MAB377)
Validation The anti-mouse NeuN was used in Figure 1a to label neurons relative to Emx1-cre positive cells. This monoclonal antibody clone
has been validated for specificity in numerous publications (2000+, see https://www.citeab.com/antibodies/226230-mab377-
anti-neun-antibody-clone-a60/publications).
3
natureresearch|reportingsummaryApril2018
Animals and other organisms
Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research
Laboratory animals All animals used in this study (Emx1-cre, RCL-eNpHR3.0/YFP, Nr5a1-cre, PV-cre, and RCL-ChR2/YFP) were maintained on a C57
background. Animal ages ranged from P40 to P250 (average P111 +/- 50 days standard deviation) at the start of training. 65% of
the subjects were female; 35% were male.
Wild animals No wild animals were involved in the study.
Field-collected samples No field-collected samples were used in the study.