Whisker sensory system – From receptor to decision
Mathew E. Diamond a,1,
*, Ehsan Arabzadeh b,2
a
Cognitive Neuroscience Sector, International School for Advanced Studies, Trieste, Italy
b
School of Psychology, University of New South Wales, Sydney 2052, New South Wales, Australia
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2. Whisker-mediated sensation as an ‘‘expert’’ capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3. Anatomical and functional organization of the pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.1. Whisker and follicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2. The ascending pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3. Connections of barrel cortex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4. Behavioral measures of tactile sensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.1. Quantification of behavioral performance: go: no-go discrimination task. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2. Quantification of behavioral performance: two-alternative forced-choice design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5. Modes of operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.1. Active sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.2. Generative and receptive modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Progress in Neurobiology 103 (2013) 28–40
A R T I C L E I N F O
Article history:
Received 15 February 2012
Received in revised form 11 May 2012
Accepted 15 May 2012
Available online 6 June 2012
Keywords:
Vibrissa
Sensory coding
Perception
Tactile
Texture
Vibration
Sensory cortex
A B S T R A C T
One of the great challenges of systems neuroscience is to understand how the neocortex transforms
neuronal representations of the physical characteristics of sensory stimuli into the percepts which can
guide the animal’s decisions. Here we present progress made in understanding behavioral and
neurophysiological aspects of a highly efficient sensory apparatus, the rat whisker system. Beginning
with the 1970s discovery of ‘‘barrels’’ in the rat and mouse brain, one line of research has focused on
unraveling the circuits that transmit information from the whiskers to the sensory cortex, together with
the cellular mechanisms that underlie sensory responses. A second, more recent line of research has
focused on tactile psychophysics, that is, quantification of the behavioral capacities supported by
whisker sensation. The opportunity to join these two lines of investigation makes whisker-mediated
sensation an exciting platform for the study of the neuronal bases of perception and decision-making.
Even more appealing is the beginning-to-end prospective offered by this system: the inquiry can start at
the level of the sensory receptor and conclude with the animal’s choice. We argue that rats can switch
between two modes of operation of the whisker sensory system: (1) generative mode and (2) receptive
mode. In the generative mode, the rat moves its whiskers forward and backward to actively seek contact
with objects and to palpate the object after initial contact. In the receptive mode, the rat immobilizes its
whiskers to optimize the collection of signals from an object that is moving by its own power. We
describe behavioral tasks that rats perform in these different modes. Next, we explore which neuronal
codes in sensory cortex account for the rats’ discrimination capacities. Finally, we present hypotheses for
mechanisms through which ‘‘downstream’’ brain regions may read out the activity of sensory cortex in
order to extract the significance of sensory stimuli and, ultimately, to select the appropriate action.
ß 2012 Elsevier Ltd. All rights reserved.
Abbreviations: A, amplitude; CA1, Cornu Ammonis area 1; ERH, entorhinal cortex; f, frequency; FC, dorsal medial prefrontal cortex; Hi, hippocampus; ms, millisecond; SI,
primary somatosensory cortex; SII, secondary somatosensory cortex; VI, primary visual cortex; PER, perirhinal cortex.
* Corresponding author. Tel.: +39 040 3787236; fax: +39 040 3787549.
E-mail addresses: diamond@sissa.it (M.E. Diamond), ehsan@unsw.edu.au (E. Arabzadeh).
1
The authors contributed equally.
2
Tel.: +61 02 9385 3523; fax: +61 02 9385 3641.
Contents lists available at SciVerse ScienceDirect
Progress in Neurobiology
journal homepage: www.elsevier.com/locate/pneurobio
0301-0082/$ – see front matter ß 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.pneurobio.2012.05.013
6. Receptor to decision in the generative mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.1. Generative mode: perceptual capacities and characteristics of the behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.2. Whisker kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6.3. Neuronal responses to texture-induced whisker motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6.4. Decoding – trial to trial read out of the neuronal signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6.5. Object invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
7. Receptor to decision in the receptive mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
7.1. Perceiving a sinusoidal vibration: behavioral capacities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.2. Perceiving a noisy vibration: behavioral capacities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7.3. Neuronal responses to whisker motion in the receptive mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
8. Beyond sensory cortex. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
8.1. Pathways for intracortical processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
8.2. Representation of touch at the end of the sensory pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
8.3. Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1. Introduction
Our goal here is to characterize a chain of events that occurs
when a rat acts upon signals received through the whiskers. We
address four questions. (i) What are the behavioral capacities
supported by the whiskers? (ii) How does whisker motion specify
the external stimulus? (iii) What neuronal codes are present in
sensory cortex? (iv) How is sensory information transformed in
subsequent stages of processing?
Why the focus on cortical processing? In the late 19th century,
Hermann Munk proposed that the cerebral cortex is responsible for
the difference between seeing elementary forms and perceiving
objects (Munk, 1881). His subjects were dogs that received either a
lesion restricted to the posterior pole of the occipital lobe or else a
lesion elsewhere, including regions farther anterior and lateral
(angular gyrus). Those with bilateral occipital lobe ablation
showed complete blindness, bumping into tables and walls. Those
with more anterior lesions, sparing the occipital pole, showed what
Munk called ‘‘psychic blindness’’ – they did not collide with
furniture, yet they did not recognize by vision previously familiar
objects.
By the late 20th century, behavioral methods had become more
precise and quantitative. We take the ideas expressed by
Whitfield (1979) as a conceptual framework. After analyzing
the behavioral effects of lesions in the auditory system, Whitfield
noted that animals can perform fine sensory discriminations even
after ablation of sensory cortex, provided the task does not require
them to transform ‘‘sensory data’’ into ‘‘objects.’’ For instance, a
cat with its auditory pathway ablated above the level of the brain
stem can localize sound; it can be trained to lick when a sound is
presented to its right, and to inhibit licking when a sound is
presented to its left. Thus, the brain stem can transmit left/right
differences in neuronal firing pattern to the centers that control
licking. But the same decorticate animal cannot be trained to
approach a sound source, once localized, on the other side of the
room (Neff and Diamond, 1958). Without cortex, the acoustic
waves are accessed only as a neuronal activity pattern within the
brain, not as a sound emanating from somewhere in the
surroundings. Extending this notion beyond the auditory system,
Whitfield postulated that even with sensory cortex ablated,
animals can act on the information present in subcortical centers
provided the task can be solved by reading out the elemental
physical characteristics of a stimulus (tone, wavelength, vibration
frequency). A deficit appears when the animal is required to
endow simple sensations with the quality of belonging to objects.
Whitfield concluded, much like Munk, that the cortex transforms
physical characteristics into the percept of real things that are
‘‘out there’’ in the world (p. 146).
A second function is implicit in the essay of Whitfield; the
cortex is critical for the storage and recall of previous sensory
experiences. The neuronal activity that encodes elemental sensory
data can gain meaning only when it is integrated with memories of
previous encounters with the same or different stimuli. Many
behaviors require sensory information to be retained, whether in
long term or short term (working) memory. Whereas neuronal
activity in the ascending pathways to cortex and in primary
sensory cortex itself subsides rapidly when a stimulus is removed,
later stages of cortex seem to have a special capacity for retaining
salient information (see Romo and de Lafuente, 2012). Recently,
this second fundamental function of neocortex has begun to be
studied in the tactile modality in rats and we will highlight some
novel findings.
2. Whisker-mediated sensation as an ‘‘expert’’ capacity
Understanding how the neocortex transforms physical characteristics
into the percept of real things that are ‘‘out there’’ in the
world has long been a challenge. A productive approach has been
to investigate ‘‘expert’’ cortical processing systems, ones that
accomplish complex transformations in a fast and reliable manner.
The efficiency of the primate visual system in extracting meaning
from visual scenes is well-known. For instance, in a task where
subjects must decide whether a briefly flashed photograph of a
natural scene contains a target category such as an animal or food,
monkeys can accurately respond as early as 160 ms after stimulus
presentation, and humans around 220 ms (Thorpe et al., 1996).
While the phenomenon of fast, precise perception can be
convincingly shown in the visual system, the mechanisms are
hard to unravel. The neuronal representation of simple features is
not completely clear even in VI and the large number of
dimensions in the stimulus space makes it difficult to quantify
neuronal selectivity to higher-order features (Yamane et al., 2008).
Even when the stimulus features that drive a neuron can be
defined, the workings underlying such selectivity occupy the realm
of abstract modeling (Kouh and Poggio, 2008). A mechanistic
account for visual object recognition remains beyond the grasp of
contemporary cognitive neuroscience, but building a comprehensive
characterization of the neuronal basis of behavior in simpler
sensory systems might be possible.
Mice and rats were adopted as laboratory animals for reasons
having little to do with integrative neuroscience, but we now know
that they, too, possess ‘‘expert’’ sensory processing systems. In
nature, they are active in dark environments and have poor vision;
their survival depends on the sense of touch. A classic study in
1912 illustrated that a rat’s ability to navigate through a raised
labyrinth depends on the use of its whiskers (Vincent, 1912).
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–40 29
Modern research has shown that whisker touch (along with
olfaction) represents the major channel through which rodents
collect information from the nearby environment (Diamond et al.,
2008c). They use their whiskers to recognize the positions of floors,
walls and objects, particularly in dark surroundings.
Are the sensory systems of rodents suitable for the study of
perceptual mechanisms? Until a few years ago, many neuroscientists
would readily attribute perception to primates but not to
rodents. This has changed as investigators have found that rodents
can be trained to weigh sensory evidence (Kepecs et al., 2008), to
assess reward statistics, to express their level of confidence in the
outcome of their choices (Lavan et al., 2011), and even to generalize
rules (Murphy et al., 2008), all in a primate-like manner. Rats
spontaneously recognize views that differ by angle, size, and
position as being instances of the same object (Tafazoli et al., 2012;
Zoccolan et al., 2009); such generalization is a hallmark of true
visual perception, and was once believed to belong only to
primates. All the work cited above indicates that the rodent brain
processes physical signals in order to build up representations of
objects and things that are ‘‘out there’’ in the world, exactly the
operation that Whitfield assigned to intracortical processing
(Whitfield, 1979).
3. Anatomical and functional organization of the pathway
Inspection of the rat’s snout reveals the grid-like layout of about
35 long and thick facial hairs known as vibrissae or whiskers
(Fig. 1A). These constitute an array of highly sensitive detectors
that project outwards and forwards from the snout to generate and
collect tactile information. The sensory pathway passes through
the brain stem and thalamus before reaching the primary
somatosensory cortex (Fig. 1B). In this section we provide a brief
overview of this exquisite system.
3.1. Whisker and follicle
Whiskers are hollow, tapered shafts; the cuticle of the whisker
consists of flat scales, overlapping like roofing slates (Voges et al.,
2012; Williams and Kramer, 2010). Another characteristic of
whiskers that differentiates them from ordinary hairs is the large
follicle, densely populated with various types of nerve endings
(Diamond, 2010; Ebara et al., 2002). Whisker motion transmits
mechanical energy to the follicle (Birdwell et al., 2007) which is
transduced into trains of action potentials by sensory receptors –
the terminals of trigeminal ganglion cells. Follicles are arranged in
five horizontal rows (A to E). There are 4 follicles in rows A and B,
and 9–12 follicles in rows C, D and E. All follicles of row A and B and
the first 7–8 follicles of rows C to E contain big whiskers also
known as macrovibrissae (Brecht et al., 1997). Each whisker is
identified by a unique letter-number combination corresponding
to its row and arc (e.g., row D, arc 2, or D2).
The vibrissa follicle (Fig. 2) is populated by receptors with
assorted morphologies and locations (Ebara et al., 2002; Rice et al.,
1986). Among the most prominent are Merkel endings. Other
populations include lanceloate endings, which are a form of free
nerve ending. The relations between the morphology and location
of a receptor and detailed neuronal response properties remain
unknown; to date, ganglion cell responses have been studied
without knowledge of the cell’s terminal structure. It is known that
many neurons in the trigeminal ganglion are sensitive to features
Fig. 1. (A) Close-up of a Wistar rat as it explores objects using its whiskers. Photograph courtesy of Mehdi Adibi. (B) Arrangement of the barrels in the left somatosensory
cortex of a rat, with each barrel labeled by its corresponding whisker. Whiskers of the D row are shown full length with their corresponding barrels highlighted in the cortical
map.
Fig. 2. Schematic view of the whisker follicle of a rat or mouse. Nerve terminations
enter through the superficial vibrissal nerve and the deep vibrissal nerve to occupy
different locations within the follicle, and their positioning is likely to be closely
related to type of hair movement that excites them (vibration, bending, pulling,
etc.). Picture courtesy of Frank Rice.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–4030
of whisker motion, such as velocity and acceleration (Arabzadeh
et al., 2005; Jones et al., 2004; Shoykhet et al., 2000). Other
ganglion cells are slowly adapting and appear suited to encode
whisker position (Lichtenstein et al., 1990; Shoykhet et al.,
2000).
Recently, Mitchinson and colleagues (Mitchinson et al., 2004,
2008), followed by Lottem and Azouz (2011), proposed mechanical
and mathematical models of transduction in the whisker follicle.
The most recent of these models is notable because it uses a single
parameter that determines the time course of the interaction
between whisker and receptor. In spite of the complex anatomical
structure and the variety of receptor types that exist within the
vibrissa follicle, this model successfully predicted the responses of
sensory receptor neurons to a number of complex tactile stimuli
(Lottem and Azouz, 2011).
3.2. The ascending pathway
Trigeminal ganglion cells emit a process that divides near the
cell body to form a peripheral branch and a central branch. The
sensory receptor endings described above are the terminals of
the peripheral branch (see details in Nicholls et al., 2012). About
200 ganglion cells innervate each whisker’s follicle (Clarke and
Bowsher, 1962; Do¨rfl, 1985). The central branch enters the brain
stem to form synapses in the trigeminal nuclei (Clarke and
Bowsher, 1962; Torvik, 1956). The trigeminal nuclei convey
afferent vibrissal information to the thalamus via parallel
pathways which then continue to the somatosensory cortex
(Descheˆnes et al., 2005). Somatosensory cortex, defined as
the area receiving direct input from the ascending somatosensory
pathway, consists of a primary field (SI) and a secondary
field (SII).
The primary field, SI, has been studied intensively in rats and
mice. In this area, macrovibrissae have a distinct representation.
Both histological (Woolsey and van der Loos, 1970) and
electrophysiological (Welker and Woolsey, 1974) studies demonstrated
a one-to-one correspondence between macrovibrissae
and barrels – distinct clusters of neurons in SI. Hence, the whiskerreceiving
area of SI is often called barrel cortex. In addition to the
wealth of knowledge provided to developmental neurobiology
(Andres and Van der Loos, 1985), the elegant topography of the
sensory pathway offers a great convenience to behavioral
neurophysiology: by simultaneous recording of barrel cortical
activity and video-monitoring of the whiskers, it is possible to
directly correlate the motion of an identified whisker with the
firing of the cortical neurons that receive input from that whisker.
Later sections will illustrate some of the conclusions available
from such correlations.
3.3. Connections of barrel cortex
Among the major subcortical targets of the somatosensory
cortex is the striatum, which contains anatomical and functional
maps of the whiskers (Alloway et al., 2000; Wright et al., 2001).
Other descending projections target the same thalamic structures
which provide sensory input (Chmielowska et al., 1989).
SI and SII send and receive dense reciprocal connections
(Carvell and Simons, 1987; Kim and Ebner, 1999). It is an open
question as to whether SI and SII in rodents function in a
hierarchical manner as is believed to be the case in primates (Pons
et al., 1992) or operate in parallel on different sorts of
somatosensory information. The functional properties of the
secondary field, SII, have been examined rarely, and only in
anesthetized animals (Carvell and Simons, 1986; Kwegyir-Afful
and Keller, 2004).
4. Behavioral measures of tactile sensation
The principal function of the nervous system is to generate
behavior. Indeed, one might argue that it is the only function of the
nervous system inasmuch as the taking of actions in accordance
with changing demands of the environment allows the organism to
survive and reproduce. Over the last decades, new methods have
been developed for the characterization of neuronal activity at the
level of single cells and neuronal ensembles. More recently,
neuroscience has, in a way, returned to its roots by exercising a
careful attention to animal behavior (an approach that was
employed long before cellular and molecular methods existed). A
particularly exciting strategy is to connect the two methods, that
is, to relate a detailed and quantitative characterization of animal
behavior to the underlying cellular and molecular mechanisms at
work in the brain. The efficiency and the accuracy of expert sensory
systems provide a good setting for such investigations.
How does the brain use information collected by sensory
systems to make decisions and select the appropriate action? We
describe two methods for quantifying under controlled laboratory
conditions how precisely a rat can distinguish between discrete
sensory stimuli.
4.1. Quantification of behavioral performance: go: no-go
discrimination task
In this procedure rats can be alternately exposed to two points
on a stimulus dimension, one of which (denoted S+) signals that
the animal’s response will produce a valuable outcome (e.g., a
sweet taste) while the other (denoted SÀ) signals that a response
will produce a negative outcome (e.g., a bitter taste; see Fig. 3A).
Rats easily solve such problems: they quickly come to respond in
the presence of S+ and refrain from responding in the presence of
SÀ. However, the simplicity of the procedure masks the complexity
of the computations required. The solution requires subjects to
parse the environment into those aspects which are relevant (the
stimulus dimension) and those which are irrelevant (for example,
common background cues), in essence, distinguishing salient from
non-salient cues. They must discriminate between the two points
on the relevant dimension with respect to their positive (S+) and
negative (SÀ) consequences, attributing appropriate motivational
significance to each cue. Variations of this basic discrimination
procedure can be used to study the accuracy with which an animal
can select and use sensory cues to optimize the outcome.
4.2. Quantification of behavioral performance: two-alternative
forced-choice design
In a go: no-go design like that outlined above, on every trial the
animal chooses between licking the reward spout (go response) or
refraining from doing so (no-go response) based on its allocation of
the stimulus to the category S+ versus SÀ. A different form of
experiment entails discrimination between pairs of stimuli
arranged into a two-alternative forced-choice design so that, on
every trial, the animal chooses between two available responses
(Fig. 3B). For example, the S+ and SÀ stimulus may be presented
simultaneously on the left and right side of the rat with the
position occupied by each changing from trial to trial. The rat
samples both stimuli and turns toward the reward spout on the
same side as the S+. In this case, the rat may receive a reward on
every trial, provided it selects the correct reward spout.
Requiring the animal to explicitly discriminate between two
stimuli provides two key advantages over a go: no-go paradigm.
First, it avoids the stimulus generalization problem that is inherent
to the go: no-go paradigm. If the animal recognizes the S+ as the
rewarded stimulus, it may generalize its positive response to SÀ
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–40 31
stimuli that it can in fact recognize as distinct from the S+. How
‘‘broadly’’ the rat will generalize from S+ depends on its criterion
for ‘‘sameness’’, its motivation, the value of the reward and other
factors which the experimenter cannot measure. To further
complicate matters, the criterion (degree of generalization) may
shift across a test session, for example as the rat approaches
satiety. All of these factors confound the experimenter’s estimate
of the animal’s capacity to discriminate S+ from SÀ. In contrast, in
the two-alternative forced-choice design the rat does not judge a
single stimulus as S+ (or close enough to S+ to warrant the same
reaction); rather, it must identify which of two stimuli is the S+.
The complication of the threshold for ‘‘sameness’’ is no longer at
play.
The second advantage is that the two-alternative forced-choice
design requires the rat to treat stimuli as events in external space.
Recall, from the discussion of Whitfield (1979), that decorticate
cats can suppress licking (no-go) when an acoustic stimulus
changes location in space, but the same cats cannot be trained to
move towards stimuli in space. Closer to the present focus on
tactile perception, Hutson and Masterton (1986) demonstrated
that after ablation of barrel cortex rats could suppress licking when
the frequency of whisker air-puff changed, but could not be trained
to jump onto a reward-platform which they localized using their
whiskers. We suggest that rats may be able to perform go: no-go
(or lick: no-lick) tasks in the absence of cortex based on the direct
transfer of information from subcortical centers to motor networks,
but interacting with the environment when cued by stimuli,
as in the two-alternative forced-choice design, is more likely to
involve a contribution of the cortex. Behaviors which require the
animal to interact with surrounding space are ideal for probing the
unique functions of cortical sensory processing. A further
advantage of the two-alternative forced-choice paradigm is that
it is more likely to involve sensory decisions in the context of a
goal-directed action. This is in contrast with the more habitual
nature of a licking response that is expected to involve reflexive
behaviors that are therefore more prone to impulsivity.
5. Modes of operation
5.1. Active sensing
Active sensing systems are purposive and information-seeking
(Prescott et al., 2011). Active sensing entails control of the sensor
apparatus, in whatever manner best suits the task, so as to
maximize information gain. Although the concept of sensor
apparatus control applies to all modalities, it is perhaps most
evident in the modality of touch.
A discussion of human tactile perception is beyond the scope of
this article, but it is interesting to note some features that
distinguish hand-mediated from whisker-mediated tactile perception,
as well as features in common. Humans (and other
primates) grasp and manipulate objects with their hands whereas
rodents do not grasp or manipulate objects with their whiskers.
(However, tactile information collected through the whiskers may
be a precursor to grasping with the paw or mouth.) Moreover,
human haptic perception relies to a great extent on proprioceptive
signals from the joints and tendons. Proprioceptive signals of this
sort are not present in the whisker follicles.
Common to human and rat tactile perception, we argue, is
variation in the mode of operation according to the ongoing task.
Humans adopt a broad range of sensorimotor strategies to collect
information through the hands. These many regimes of acquisition
Fig. 3. (A) An example of a go: no-go design also known as the ‘‘licking suppression’’ paradigm. The rat is required to refrain from licking for a specific period of time (no
licking) after which the stimulus is presented. The rat is then given a response window during which it decides whether to lick the spout or to suppress licking. The rat is
normally head-fixed. Licks that are generated in response to the stimulus designated as S+ are rewarded whereas licks to SÀ are unrewarded (or punished either by bitter
taste, by a loud sound, by a shock, or by further delaying the next stimulus presentation with a time-out penalty). (B) Two-alternative forced-choice paradigm. The rat initiates
a trial by a nose poke. After a delay period the rat receives two simultaneous stimuli on its left and right whisker pads. One of the two stimuli is designated as S+ and the other
as SÀ. The rat is required to explicitly make a discrimination and to identify the S+ and turn toward the corresponding drinking spout. Correct choices are rewarded.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–4032
are collectively referred to as ‘‘haptic exploration’’ (Lederman and
Klatzky, 1987). They range from following edges, palpating
surfaces to detect texture and softness, and resting the fingertips
on an object to detect vibration or motion (Jones and Lederman,
2006).
The rat whisker-mediated sensory system is also a prominent
case of active sensing inasmuch as the rat precisely controls its
whiskers. Self-generated whisker motion is critical for wall
following (Jenks et al., 2010), distance estimation (Harris et al.,
1999), and identifying properties such as texture (Diamond et al.,
2008a), shape and size (Brecht et al., 1997; Harvey et al., 2001). As a
rat or mouse feels its way through the world, it senses its own
whisking (Ganguly and Kleinfeld, 2004). From the relationship
between the whisking cycle and the contact signal (Curtis and
Kleinfeld, 2009) the animal localizes objects with millimeterprecision
(Knutsen et al., 2006). The extraordinary sensitivity of
receptors in the follicle allows the rat to detect tiny whisker
deflections (Adibi and Arabzadeh, 2011; Stuttgen and Schwarz,
2008).
5.2. Generative and receptive modes
Rather than review each of these, we wish to simplify the full
complement of sensory capacities by arguing that whiskermediated
perception can arise through two general modes of
operation: (1) generative mode and (2) receptive mode. In the
generative mode, the rat moves its whiskers forward and backward
to actively seek contact with objects and to palpate the object after
initial contact. The animal causes the percept by its own motion.
We will focus on the discrimination of texture as a condition in
which rats generate neuronal sensory representations through
their own whisker motion. In the receptive mode, rats immobilize
their whiskers to optimize the collection of signals from an object
that is moving by its own power. We will focus on the
discrimination of vibrations applied to the whiskers by external
devices as an operation in the receptive mode.
We shall raise a number of questions. How do rats approach the
object to be identified? How do the whiskers transduce object
properties? How is a continuous stimulus dimension partitioned
into behaviorally relevant categories and how are such categories
encoded in memory? We first describe behavioral paradigms that
allow quantification of performance for both generative and
receptive modes of operation of the whisker sensory system. Next,
we will outline the neuronal processing properties that underlie
these behaviors. Finally, we propose a model for how an intrinsic
stimulus representation is modified by discrimination learning to
generate behaviorally relevant categories to guide action.
6. Receptor to decision in the generative mode
The generative mode of sensing in humans was described by the
philosopher Merleau-Ponty:
[...] the knowing touch projects us outside our body through
movement. [...] There are tactile phenomena, alleged tactile
qualities, like roughness and smoothness, which disappear
completely if the exploratory movement is eliminated.
Movement and time are not only an objective condition of
knowing touch, but a phenomenal component of tactile data.
They bring about the patterning of tactile phenomena, just as
light shows up the configuration of a visible surface. (MerleauPonty,
1945) (p. 367 of translation by Colin Smith.)
Rats, in the generative mode, sweep their whiskers forwards
and backwards at a frequency of about 10 Hz, describing elegant
arcs (Berg and Kleinfeld, 2003; Brecht et al., 1997; Carvell and
Simons, 1990; Hill et al., 2008; Perkon et al., 2011; Zuo et al., 2011).
The rhythmic sweeping action is called ‘‘whisking’’ and an
individual cycle is called a ‘‘whisk.’’ Whisking is the starting point
for the perception of texture, as described in the following sections.
6.1. Generative mode: perceptual capacities and characteristics of the
behavior
The perception of surface texture is critical to rodent behavior,
for example in the selection of nesting materials (Rhodes and
Richmond, 1985). A number of tasks have been devised to test
texture perception. For example, when rats (in the dark) face two
platforms with different textures at the entryway, they can learn to
identify the reward-associated texture and jump to the correct
platform. In tasks like this, rats learn to extract the identity of a
texture based on one to three touches per whisker and display
accurate judgments of a texture within 100 ms of initial whisker
contact (von Heimendahl et al., 2007). Rats show high discriminative
capacities when they are trained to compare two spatial
densities of grooves on a surface (Carvell and Simons, 1990). More
recently, Morita and colleagues (Morita et al., 2011) trained rats to
discriminate between sandpapers of various degrees of roughness.
Using their whiskers, rats could successfully discriminate two
sandpapers with a difference in mean grit size of just 100 versus
82 mm. The accuracy of their judgments (Morita et al., 2011),
combined with the speed of decision making (von Heimendahl
et al., 2007) indicates a high level of efficiency. How does the brain
accomplish whisker-mediated texture discrimination?
For an efficient decoding of the sensory signal arising in the
generative mode, the sensory system must integrate ‘‘knowledge’’
of motor output in order to correctly interpret incoming
information. Just as the brain would not be able to estimate the
weight of an object we are lifting without taking into account the
motor signals that produce muscle contraction, the afferent signal
from a whisker cannot be optimally decoded without information
about the whisker movement that generated the tactile signal.
How does the sensory system obtain knowledge of the executed
movement? There are two possibilities. First, sensory pathways
might receive copies of the motor signal from the brainstem, zona
incerta, the cerebellum and the motor cortices (Ahrens and
Kleinfeld, 2004; Veinante and Deschenes, 2003). Second, the
sensory pathway itself might carry afferent signals about whisking
(Yu et al., 2006). Whichever of the two mechanisms is at play (or,
more likely, some combination), all available anatomical and
physiological evidence indicates that the barrel cortex is a direct
participant in the motor network.
To identify surface texture rats make a sequence of brief
contacts with their whiskers (von Heimendahl et al., 2007).
Consistent with its fine motor control, whisking has been shown to
be tuned to task requirements (Berg and Kleinfeld, 2003; Carvell
and Simons, 1990; Harvey et al., 2001). Performance in a texture
discrimination task was found to be positively correlated with the
frequency bandwidth and the duration of whisking (Carvell and
Simons, 1995). The use of high-speed video cameras has allowed
whisker kinematics to be studied at higher temporal and spatial
resolutions as whiskers come in contact with textures in an awake
behaving rat. A recent study found evidence for the use of an
adaptive motor strategy in the whisker system. In a two alternative
forced-choice paradigm, rats were trained to classify surfaces
based on the specific distancing of their grooves. When they were
subjected to the clipping of the whiskers, leaving about 5 per side
as compared to the normal complement of more than 30 per side,
their behavior changed in ways that appear consistent with an
optimization of motor output to compensate for lost information.
The rats increased the total contact time, per trial, of the remaining
whiskers. These results lead to the intriguing notion that the rats
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–40 33
use an information-seeking ‘cognitive’ motor strategy, instead of a
rigid motor program (Zuo et al., 2011).
6.2. Whisker kinematics
Since whisker motion is the starting point for any tactile
capacity, a critical step is to understand how motion represents the
contacted surface. During lateral motion along textured surfaces,
surface features such as groove depth or grain size, and the spacing
between them, would be expected to modulate the movement
profile. Can one identify the contacted surface based on the
observation of the movement profile? The whisker-texture contact
profile was firstly characterized at high temporal and spatial
resolutions in anesthetized preparation using electrical stimulation
of the facial nerve – electrical whisking (Arabzadeh et al., 2005).
Various textures were then placed in contact with the whisker to
generate a texture library – a collection of movement profiles at the
base of the whisker for hundreds of sweeps over each texture.
Dissection of the kinetic profiles revealed that as a whisker
palpates a texture, its movement transiently changes with every
contact and release from surface grains – the characteristic highvelocity,
high-acceleration event that arises from the whiskergrain
interaction is also known as a stick-slip event. Each texture
resulted in a unique ‘‘kinetic signature’’ defined by the number and
the profile of the high-velocity events (Arabzadeh et al., 2005).
Further experiments recorded whisker motion in awake rats as
they whisked against sandpapers of different grain size and
confirmed that the rate and magnitude of kinetic events varied
systematically with sandpaper textures (Wolfe et al., 2008) as well
as grooved surfaces (Zuo et al., 2011). The whisker movement
profiles obtained during awake active whisking against sandpapers
were remarkably similar to those collected during electrical
whisking (Diamond et al., 2008a). Fig. 4 shows an example of a slip
movement as the whisker springs loose in posterior direction after
it is released from a groove during a texture classification task (Zuo
et al., 2011).
6.3. Neuronal responses to texture-induced whisker motion
The next step was to verify whether whisker motion profiles
unique to each texture – the ‘‘kinetic signatures’’ of surfaces –
produce reliable and systematically distinct neuronal responses.
For this hypothesis to be viable, the kinetic signatures must fulfill
two criteria: (i) they must evoke neuronal activity that carries
information about texture and (ii) the neuronal responses must be
available in the time interval during which the behavioral
discrimination is performed and must directly influence the
animal’s choices (Diamond et al., 2008b).
Recording neuronal response to various textured stimuli in an
anesthetized preparation confirmed the first criterion. The texture
library was replayed to the base of the whiskers of anesthetized
rats while neuronal responses were obtained at two levels of the
pathway. Responses of the first-order neurons of the trigeminal
ganglion and those of barrel cortex showed a reliable encoding of
the kinetic signature – at every stage, individual neurons as well as
neuronal ensembles were effectively driven by the high-velocity
whisker movements and thus replicated the kinetic signature in
their response profile (Arabzadeh et al., 2005). Lottem and Azouz
(2008) induced electrical whisking in anesthetized rats and
concurrently recorded from trigeminal ganglion neurons. Again,
despite the high variability in whisking parameters, different
textures were translated into distinct movement profiles and
resulted in distinct neuronal responses in the first-order neurons.
Testing the second criterion requires measurements in behaving
animals. Simultaneous recording of cortical neuronal activity
from rats performing texture discrimination tasks confirmed the
predictions of the anesthetized studies (Jadhav et al., 2009; von
Heimendahl et al., 2007). Alignment of high energy stick-slip
events with neuronal responding in barrel cortex revealed a
transient increase in firing rate and firing synchrony on surfaces
(Jadhav et al., 2009).
Currently two biologically plausible coding mechanisms have
been identified, the ‘‘temporal integration’’ and the ‘‘temporal
pattern’’ hypotheses. According to the first hypothesis, the brain
encodes texture by a single quantity of spikes (total number of
spikes, or else spikes per unit of time) accumulated across a whisk.
In short, when the texture-specific kinetic signature causes high
energy (i.e., high velocity and acceleration) movements to reach
the receptors in the follicle, high firing rates are evoked; lower
energy movement elicits lower firing rate. According to the second
hypothesis, textures are encoded in the temporal sequence of high
energy events within the kinetic signature. For example, one
texture may evoke a kinetic signature with regularly timed stickslip
events, and a second texture may evoke a signature with
alternating long and short intervals. The temporal pattern of
kinetic events is captured in the neuronal firing pattern (Arabzadeh
et al., 2006), and if the readout mechanism can decode firing
patterns, then the animal would possess a much higher capacity for
representing textures than if it used only the firing rate decoding
mechanism.
6.4. Decoding – trial to trial read out of the neuronal signal
Decoding refers to the process by which a ‘‘downstream’’
neuronal population extracts information contained in the firing of
an ‘‘upstream’’ neuronal population. Although the investigator can
never specify decoding mechanisms with certainty, hypotheses
can be tested by determining whether a proposed mechanism is
consistent with the animal’s behavior. More specifically, if the
information extracted through a candidate decoding mechanism
can explain the trial by trial variation in an animal’s percept
(measured by its behavioral choice) then that mechanism gains
validity as a potential decoding principle. The perception of surface
texture depends upon cortical processing (Guic-Robles et al.,
1992). For this reason, we look to cortex for the representation of
texture best correlated with the animal’s explicit judgment of the
stimulus. How are spike trains in the sensory system ‘‘read out’’ to
support a behavioral discrimination between textures? In a texture
discrimination paradigm, rats were trained to perch at the edge of a
platform to touch a textured plate with their whiskers, and then to
indicate their discrimination of the texture’s identity by turning
either left or right to collect a water reward (von Heimendahl et al.,
Fig. 4. Frame-to-frame tracking of whisker position as it gets stuck in a groove and is
subsequently released at high velocity. Whisker C4 is traced in color over sequential
1 ms steps. Whisker kinematic events like this provide input to sensory receptors
that allow the brain to identify the texture.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–4034
2007). On correct trials when rats accurately identified the stimulus,
75 ms before the rat made its choice, the average firing rate of the
cortical neurons varied according to the contacted texture. It was
higher for rough textures and lower for smooth textures. For trials of
incorrect discrimination, at 75 ms before the rat made its choice, the
firing-rate code was reversed; it was lower for rough textures and
higher for smooth textures. Thus when the barrel cortex neurons
carried the ‘‘right’’ signal the rat was likely to make a correct choice,
decoding the signal to mean surface was ‘rough’, and when they
carried the ‘‘wrong’’ signal it was likely to make an incorrect choice,
decoding the signal to mean the surface was ‘rough’, when it was
actually ‘smooth’. Therefore in each trial the rat made its decision
based on the magnitude of whisker evoked activity in the barrel
cortex. Mutual information analysis revealed that the temporal
pattern of spikes can provide extra information to the brain above
and beyond what is available in spike counts (Arabzadeh et al.,
2006). However, it is not clear whether this extra information can be
read out by the brain and thus can contribute to behavior. The
authors are currently examining larger data sets to determine
whether there is texture information present in the temporal firing
patterns of cortical populations and whether that information is
decoded, e.g., whether the temporal pattern information present
correlates with the animal’s choice.
6.5. Object invariance
In the real world, incoming sensory data are never identical in
two encounters with the same object. For instance, observations of
behavior reveal high levels of variability in the way rats sample
surface textures from one trial to another. This includes variability
in the angle of approach, the distance from snout to texture, the
amplitude, frequency, and speed of whisker motion as well as
variability in the number of whiskers that come into contact with
the surface (Zuo et al., 2011). Trial to trial differences in the way
whiskers engage a surface may cause spike patterns to vary,
making the precise temporal code less robust than a spike count
code. Yet, when the position or angle of a textured plate is changed,
a trained rat correctly identifies the texture on its first encounter
(unpublished observations). In spite of such trial to trial variability
rats achieve high levels of performance. How do cortical neurons
attain texture invariance? Currently little is known about how the
neuronal kinetic signatures survive trial to trial variability and give
rise to texture invariance.
However, one line of evidence comes from a close analysis of
the texture library and the neuronal activity recorded across the
pathway when a component of the library was replayed to the
whiskers (Arabzadeh et al., 2005). The library consisted of kinetic
data collected for hundreds of sweeps over various textures. This
allowed us to remove trial to trial variability in a subset of trials: a
specific movement profile was selected from the library and
presented repeatedly. While first-order responses were exquisitely
reliable, cortical responses showed significant variations across
these repeated trials. Because of the precise temporal alignment of
spikes to the high-velocity events, the first-order neuron response
was nearly identical on each trial – across 100 stimulus
presentations spike times corresponding to a specific high-velocity
event had a standard deviation of about 0.1 ms or less. Individual
sweeps over the same texture thus resulted in distinct response
profiles in the first-order neuron. While cortical neurons were
highly informative about texture identity, they did not differentiate
between multiple sweeps over the same texture; the cortical
response to a repeated replay was similar to their response to one
hundred unique sweeps over the same texture.
The mechanisms underlying invariant perception of textures
are yet to be investigated in behaving rats. The study will require
direct correlations between various aspects of neuronal activity
and the rats’ response.
7. Receptor to decision in the receptive mode
If we need to check whether our computer has been turned off,
we would likely place our fingertips lightly on the case to feel for
vibrations produced by the fan. It is unlikely we would palpate the
surface and sweep our fingertips along it, as we would do for a
texture judgment (Gamzu and Ahissar, 2001). Such active motion
can confound the skin vibration emanating from the computer fan
with the skin vibration produced by motion along surface features.
Thus, we (primates) adjust our hand and finger motor output
according to what information we need to extract about the objects
around us.
It is difficult to quantify rodents’ use of their whiskers in
natural, out-of-laboratory settings. But even in the absence of
objective data it seems reasonable to assume that some forms of
perception rely on blocking motor output to keep the whiskers
immobile, much the same as we block hand motion. For example,
how do rats perceive the passage of a large predator above their
burrow? We speculate that they place their whiskers in contact
with the walls and floor, with negligible whisking output, to
‘‘listen’’ for vibrations (see Fig. 5).
Fig. 5. The receptive mode. As a predator approaches the rat’s hiding place, the vibration signal might be transferred to the whiskers through their contact with the walls and
floor of the burrow. Changes in vibration intensity over short time intervals would provide important information about the speed and direction of the predator. Drawing by
Marco Gigante.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–40 35
It is tempting to name the state of the sensory system
characterized by exploratory whisking as ‘‘active’’ and the state
of quiet immobility as ‘‘passive’’ (Kleinfeld et al., 2006), but we
suggest that this nomenclature is misleading in its implication that
the nervous system itself becomes passive in the immobile state,
waiting to be subjected to unknown events. Behavioral work (see
below) indicates that the animal is highly ‘‘active’’ even when it
places and holds its whiskers in contact with a moving stimulus.
For this reason we refer to the ‘‘quiet’’ whisker and brain state as
the ‘‘receptive mode’’ rather than the passive mode.
We can further develop the illustration of the rat feeling for
ground vibrations in the receptive mode. If the burrow’s walls
tremble, is the predator approaching (increasing vibration
intensity) or moving away (decreasing vibration intensity)?
Changes and differences in vibration intensity seem ecologically
relevant. In the sections below, we describe two behaviors, and
possible neuronal correlates, that arise from judgments of whisker
vibration intensity.
7.1. Perceiving a sinusoidal vibration: behavioral capacities
Rats were trained in a behavioral paradigm (Fig. 6) involving
discrimination of vibrotactile stimuli (Adibi et al., 2012). Rats
started each trial by a nose-poke into the stimulus aperture where
whiskers came in contact with two mesh plates. After a random
delay period (during which rats were required to sustain the nosepoke
position) two vibration stimuli were presented simultaneously
on the right and the left mesh plates. To receive a reward,
rats had to turn towards the side with the higher intensity
vibration (i.e., the stimulus designated as S+). What features do rats
use to compare the intensity of two vibrations? Which features of
whisker motion does the sensory system extract to construct
percepts? We addressed this question by training rats to make
discriminations between sinusoidal vibrations that differed in
amplitude (A) or frequency (f) or both. One set of rats learned to
reliably identify which of two vibrations had higher frequency (f1
versus f2) when amplitudes were equal. Another set of rats learned
to reliably identify which of two vibrations had higher amplitude
(A1 versus A2) when frequencies were equal. While these results
indicate that both elemental features contribute to the rats’
sensation, a further test found that the capacity to discriminate A
and f was reduced to chance when the difference in one feature was
counterbalanced by the difference in the other feature: rats could
not discriminate amplitude or frequency whenever A1f1 = A2f2.
Fig. 7A shows a summary of the behavioral results. Thus, vibrations
were sensed as the product Af rather than as separable elemental
features, A and f. The product Af is proportional to a physical entity,
the mean speed. Analysis of performance revealed that rats
extracted more information about differences in Af than predicted
by the sum of the information in elemental differences. These
behavioral experiments support the predictions of earlier physiological
studies by demonstrating that rats are ‘‘blind’’ to the
elemental features present in a sinusoidal whisker vibration;
instead, they perceive a composite feature, the speed of whisker
motion.
7.2. Perceiving a noisy vibration: behavioral capacities
We recently devised a second task to measure the capacity of
rats to perceive differences in vibration intensity (Fassihi et al.,
2012). Rather than comparing two stimuli applied simultaneously
to their left- and right-side whiskers, the rats compare two stimuli
delivered sequentially. At the outset, they are trained to position
their snout in a nose hole such that their head is stationary and
their right-side whiskers are firmly in contact with a plate that
delivers motion along the anterior–posterior axis. Instead of a
perfectly repeating sinusoid (Adibi et al., 2012), the stimulus is a
sequence of position values drawn from a Gaussian distribution
with standard deviation denoted s. The first stimulus is called
‘‘base’’ (defined by s1) and the second stimulus ‘‘comparison’’
(defined by s2). In the standard protocol, the duration of base and
comparison stimuli both are 400 ms. By considering trials with s1
fixed while s2 spans a range of values, psychometric curves can be
constructed. Performance is above 75% correct when the stimulus
intensity difference index defined as the absolute value of
(s1 À s2)/(s1 + s2), is larger than 0.1. The discriminative capacity
of rats in this task is equivalent to that of human subjects who
receive the same stimulus set on their fingertip (Fassihi et al.,
2012).
It has been debated how sensory systems accumulate stimulus
information over time; it has been suggested that the maximum
time across which whisker motion could be integrated to lead to a
behavioral choice may be as low as 25 ms (Stu¨ ttgen and Schwarz,
2010). To investigate this, we varied the duration of the
comparison stimulus: 200, 400, or 600 ms. The rats’ performance
improved for longer comparison stimuli, suggesting that for
stimuli with a probabilistic structure, evidence can be accumulated
over time.
The task described above is appealing not only because it will
allow insights into how ‘‘noisy’’, unpredictable stimuli are encoded
Fig. 6. Schematic representation of the vibrotactile discrimination task [from Adibi et al., 2012]. The rat initiated each trial by a nose-poke into the stimulus aperture while
touching the two mesh plates with its whiskers (left panel). After a random delay period during which nose-poke was continually maintained, the stimuli were presented
simultaneously on the left and right mesh (middle panel). Having identified the plate vibrating with S+ the rat made a behavioral choice by turning towards the corresponding
drinking spouts (circles) and receive a reward (right panel).
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–4036
in the sensory system, but also because it is constructed in such a
way as to force the rat to retain information about the ‘‘base’’
stimulus in working memory. Although rodents can perform
spatial alternation and odor or object-guided delayed match-tosample,
there is no study showing working memory for stimuli
distributed along a sensory continuum. The main result to date is
that performance remains high across all delays up to 4 s, the
longest tested (Fassihi et al., 2012).
7.3. Neuronal responses to whisker motion in the receptive mode
The outcome of the behavioral study on perception of
sinusoidal vibrations is consistent with neuronal responses to
the same stimuli. In earlier studies, we analyzed the cortical
neuronal activity evoked by sinusoidal vibrations of the whisker
shaft of anesthetized rats. We systematically varied vibration
amplitude (A) and frequency (f) while making extracellular
recordings from barrel cortex neurons to construct the neuronal
response functions (Arabzadeh et al., 2003, 2004). The results
demonstrated that single neurons and cortical ensembles reliably
encode the product Af by their firing rate: increases in Af were
accompanied by increases in firing rate. Methods of mutual
information quantified the signal carried by barrel cortex neurons
about the two elemental features A and f separately and their
potential joint encoding. Neurons were found to reduce the
dimensionality of the stimulus from two features (A, f) to a single
feature, the product Af (Arabzadeh et al., 2003, 2004) (Fig. 7B).
Furthermore, the nature of this code was highly efficient for a
population – because different neurons encoded stimuli in the
same manner, information loss was negligible even when the
activity of separate neuronal clusters was pooled.
Importantly, the two elemental features A and f could not be
decoded separately: any given firing rate could result from an
infinite number of individual A and f values provided that they
formed, together, a single value of Af. Thus, in summary, the failure
of neuronal populations to distinguish between the two elemental
features seems to be the clearest explanation for the failure of
behaving rats to distinguish between these same two elemental
features.
Experiments using aperiodic, unpredictable stimuli provide
predictions for how sensory cortical neurons may encode the
stimuli used in the ‘‘noise’’ comparison task. Neurons in barrel
cortex are more responsive to noisy stimuli than regular, periodic
stimuli (Lak et al., 2008, 2010; Maravall et al., 2007; Petersen et al.,
2009). A preliminary analysis of neuronal activity from anesthetized
rats provides support for a simple model where (i) both firing
rate and spike count are correlated with the stimulus position
standard deviation s and (ii) differences in both firing rate and
spike count could be decoded to allow comparison of the base and
comparison stimuli (Fassihi et al., 2012). The change in performance
as a function of variable stimulus duration suggests that
firing rate is a better candidate than spike count as the decoded
feature.
8. Beyond sensory cortex
In visual perception, object identification is believed to emerge
in the ‘‘ventral’’ processing pathway that travels within the
inferotemporal lobe, passing from primary visual cortex to the
hippocampus (Devlin and Price, 2007). The study of sensory
processing in rats beyond the primary cortical fields is in its
infancy. We wish to present hypotheses and predictions, as well as
some preliminary findings. Our thinking is motivated by investigations
of visual object perception; for this reason, we suppose
that processing of tactile information generates the animal’s
knowledge of the object that it is contacting through stages of
transformation from primary sensory cortex to hippocampus.
8.1. Pathways for intracortical processing
As in the primate visual system, anatomical evidence in rats
suggests that somatosensory information reaches the hippocampus.
Fig. 8 gives a schematic view of the intracortical pathways and
the key areas involved. We will focus on the dual routes by which
tactile information from somatosensory cortex (SI plus SII) reaches
the entorhinal cortex and, from there, hippocampus (Burwell and
Amaral, 1998; Swanson, 1981).
Route I. Sensory cortex projects in the anterior direction
directly to the dorsal part of medial prefrontal cortex (FC) – SI
sparsely and SII more densely. This part of medial prefrontal
cortex is believed to be an analogue of the dorsolateral
prefrontal cortex of primates. Prefrontal areas, in turn, project
to the lateral entorhinal area and directly to hippocampus.
Route II. Sensory cortex also projects to perirhinal cortex (PER).
The target region consists of the upper bank of the rhinal sulcus
(homolog of Brodmann area 36 in primates) and the lower bank
of the rhinal sulcus (homolog of area 35). Perirhinal cortex
projects to the entorhinal cortex (ERH). Entorhinal cortex
receives input from all intracortical sensory processing streams
and is the gateway to the hippocampus proper. Medial
entorhinal area is mostly supplied via the postrhinal cortex
Fig. 7. (A) Behavioral performances across various stimulus dimensions [from Adibi et al., 2012]. Rats could discriminate between stimuli with pure differences in amplitude
(delta A) or frequency (delta f). Performance was highest when changes in amplitude and frequency were combined in the same direction (delta Af). When changes in
amplitude and frequency were combined in the opposite direction (iso Af) the resulting vibrations were indiscriminable. (B) Response magnitude function of a typical barrel
cortex neuron to sinusoidal vibration stimuli [from Arabzadeh et al., 2003, 2004]. The four types of discrimination, as defined in A, are illustrated by green lines.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–40 37
and is the established seat of grid cells (Hafting et al., 2005). In
contrast, lateral entorhinal area receives a stronger input from
perirhinal cortex (and direct, but weaker, inputs from SI and SII)
and is likely to be the relay stage for tactile inputs to the
hippocampus.
Does tactile information indeed reach the hippocampus, and, if
so, in what form? In rats, the best-described property of
hippocampal cells is the place field (O’Keefe, 1976) but evidence
is accumulating that neurons also encode events or objects in
conjunction with space (Fortin et al., 2004). This evidence for a
more general function makes rat hippocampus analogous to
human hippocampus, where neurons are active during storage and
recall of recent episodes (Gelbard-Sagiv et al., 2008).
8.2. Representation of touch at the end of the sensory pathway
We recently designed an experiment to determine whether
texture is encoded in the hippocampus (Itskov et al., 2011). The
critical step was to increase the number of stimuli from two
textures, with a single texture associated with each action of the
rat (turn left versus turn right) to four textures. Since now two
different textures were associated with the same reward location
and action, any difference in activity must reflect the coding of
touch rather than some aspect of explicit behavior (Fig. 9). In this
way our experiments were able to tease out the stimulus
representation in hippocampus without spatial or behavioral
confounds. We found that 18% of neurons sampled in CA1 region of
hippocampus discriminated between two textures associated with
the same action. Because the texture pair was presented in the
same location and the rat moved through space in the identical
way in response to contact, the result shows an explicit
hippocampal representation of touch, beyond that of space. We
interpret these findings as an indication that the hippocampus has
the function of forming episodes composed of the relationship
between tactile experience and its spatial context.
As we conduct new experiments, our working hypothesis is that
along the stream of cortical areas neurons will show a progressive
shift in the nature of information carried, from physical features
(‘‘what is the motion of the whiskers?’’) to the meaning of the
sensation (‘‘what are the whiskers touching?’’). Texture properties
will be represented in SI as a continuum: neuronal firing rate encode
the physical properties of the contacted surface such as groove
density. In later stages of processing, we expect to find two classes of
neurons, those with positive and those with negative slopes. We
speculate that the formation of dual populations beyond somatosensory
cortex is a generalization of the finding in primates that SII
neurons exhibit both positive and negative slopes for increasing
vibrotactile frequency (Romo and de Lafuente, 2012).
An appealing notion is that perception is achieved by the
abstraction of stimuli from a continuum, allowing physically
similar events to take on very different meanings. Stimuli acquire
Fig. 8. Intra-cortical streams flowing from the vibrissal somatosensory cortex
consisting of a primary (SI) and a secondary field (SII) with reciprocal connections.
Hippocampus (Hi), buried below the surface, is uncovered and moved for this
illustration. PER, perirhinal cortex; ERH, entorhinal cortex; FC represents the dorsal
part of the medial prefrontal cortex.
Fig. 9. (A) Behavioral paradigm from (Itskov et al., 2011). On each trial the rat approached the center of the platform to touch a textured plate with its whiskers. Four textures
were used (T1 to T4). The rat then discriminated the surface of the plate and, according to its identity, turned left or right to obtain a reward. (B) Top panel: firing of this CA1
neuron distinguished between T2 and T3, associated with the same action. Neuronal activity averaged across trials (s.e.m. in shading) aligned to reward collection (0 s). The
vertical dashed lines indicate the beginning of the touch (red), turn (blue) and the reward (green) phase of the behavior. Lower panel illustrates how the texture information
that is carried in the response of this neuron rises during the touch phase.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–4038
‘‘labels’’ or ‘‘tokens’’ as they are processed along intracortical
streams. In our study, the neuronal representation of textures in
the hippocampus had no relation to the graded responses found in
SI: any given neuron could give its maximum response for any of
the textured stimuli, not necessarily the roughest one (Itskov et al.,
2011). It is clear that the responses were no longer a ‘‘sensory code’’
but a representation of discrete objects.
8.3. Challenges
How do impulses from skin receptors eventually lead to the
sensation (i.e., the registration of the elemental properties) and
perception (i.e., recognition of the identity and meaning of the
sensation)? The words of Whitfield are (again) well composed:
‘‘When we hear the sound of a bell, we know perfectly well, as
men have known from antiquity, that in no way does the
vibrating bell itself enter the nervous system. Yet we become
aware of the ringing bell as an external object, and not as a
series of sense data. Nevertheless, the sound enters the ear only
as a set of aerial vibrations, and it is only this information that
the sensory transducers have to work on. It is the transformation
between the two representations that is the essence of our
problem.’’ (p.130)
To make Whitfield’s idea more explicit, consider that all the
‘‘raw data’’ that sensory receptors collect from the outside world
are present in the peripheral sensory system. At the entry regions
to the cerebral cortex – the sensory cortical areas – neuronal
activity constitutes a rich and diversified representation of the
physical events impinging on the receptors. While the specific
operations carried out on sensory signals once they reach the
neocortex depend on the sensory modality (and even within the
acoustic modality, words are processed differently than music
(Peretz, 2006)), one general answer holds up for any input –
intracortical processing serves to integrate, elaborate, and
distribute elemental sensory signals in such a way that those
signals gain new meaning through learning or, if familiar, can be
interpreted by linkage with stored knowledge. In short, the
transformations within cortex serve to construct perception – the
abstraction of stimuli from a continuum, allowing physically
similar events to take on very different meanings. Stimuli acquire
‘‘labels’’ or ‘‘tokens’’ as they are processed along intracortical
streams. Thus, when we take a fruit in hand, we feel its texture and
recognize a peach; we do not (unless asked to do so) perceive the
pattern of activation of the receptors in our skin.
Work done since Whitfield’s 1979 review indicates that, beyond
attributing meaning to sensory data, an additional and related
function of cortex is to compare current stimuli to those stored in
memory. This function comes into play whether the stimuli are realworld
things, like the peach that is being palpated, or are simpler
sensory data like a computer-controlled skin vibration. Both cases
require access to stored information – the memory of what a peach
feels like, or the memory of what the vibration delivered a few
seconds earlier felt like. Such stored representations appear to reside
in cortex. Romo and de Lafuente (2012) emphasize the memory
operations of primate cortex, and we expect that in the whiskermediated
working memory task, analogous operations will also be
discovered in the rat cortical processing streams. A further step in
cortical processing is to partition a continuous stimulus dimension
into behaviourally relevant categories and to assign predictive
values to stimulus categories based on their associative strength
within an environment and the uncertainty with which they predict
rewards. It is safe to say that the whisker sensory system promises
fertile ground for the continuing study of the transformation from
receptor to decision.
Acknowledgments
This work was supported by the Human Frontier Science
Program (contract RG0041/2009-C), the European Research
Council, the European Union (contract BIOTACT-21590 and
CORONET), the Compagnia San Paolo, the Australian Research
Council Discovery Project DP0987133 and Australian National
Health & Medical Research Council Project Grant 1028670.
References
Adibi, M., Arabzadeh, E., 2011. A comparison of neuronal and behavioral detection
and discrimination performances in rat whisker system. Journal of Neurophysiology
105, 356–365.
Adibi, M., Diamond, M.E., Arabzadeh, E., 2012. Behavioral study of whisker-mediated
vibration sensation in rats. Proceedings of the National Academy of
Sciences of the United States of America 109, 971–976.
Ahrens, K.F., Kleinfeld, D., 2004. Current flow in vibrissa motor cortex can phaselock
with exploratory rhythmic whisking in rat. Journal of Neurophysiology 92,
1700–1707.
Alloway, K.D., Mutic, J.J., Hoffer, Z.S., Hoover, J.E., 2000. Overlapping corticostriatal
projections from the rodent vibrissal representations in primary and secondary
somatosensory cortex. Journal of Comparative Neurology 428, 51–67.
Andres, F.L., Van der Loos, H., 1985. From sensory periphery to cortex: the architecture
of the barrelfield as modified by various early manipulations of the
mouse whiskerpad. Anatomy and Embryology 172, 11–20.
Arabzadeh, E., Panzeri, S., Diamond, M.E., 2004. Whisker vibration information
carried by rat barrel cortex neurons. Journal of Neuroscience 24, 6011–6020.
Arabzadeh, E., Panzeri, S., Diamond, M.E., 2006. Deciphering the spike train of a
sensory neuron: counts and temporal patterns in the rat whisker pathway.
Journal of Neuroscience 26, 9216–9226.
Arabzadeh, E., Petersen, R.S., Diamond, M.E., 2003. Encoding of whisker vibration by
rat barrel cortex neurons: implications for texture discrimination. Journal of
Neuroscience 23, 9146–9154.
Arabzadeh, E., Zorzin, E., Diamond, M.E., 2005. Neuronal encoding of texture in the
whisker sensory pathway. PLoS Biology 3, e17.
Berg, R.W., Kleinfeld, D., 2003. Rhythmic whisking by rat: retraction as well as
protraction of the vibrissae is under active muscular control. Journal of Neurophysiology
89, 104–117.
Birdwell, J.A., Solomon, J.H., Thajchayapong, M., Taylor, M.A., Cheely, M., Towal, R.B.,
Conradt, J., Hartmann, M.J., 2007. Biomechanical models for radial distance determination
by the rat vibrissal system. Journal of Neurophysiology 98, 2439–2455.
Brecht, M., Preilowski, B., Merzenich, M.M., 1997. Functional architecture of the
mystacial vibrissae. Behavioural Brain Research 84, 81–97.
Burwell, R.D., Amaral, D.G., 1998. Cortical afferents of the perirhinal, postrhinal, and
entorhinal cortices of the rat. Journal of Comparative Neurology 398, 179–205.
Carvell, G.E., Simons, D.J., 1986. Somatotopic organization of the second somatosensory
area (SII) in the cerebral cortex of the mouse. Somatosensory Research
3, 213–237.
Carvell, G.E., Simons, D.J., 1987. Thalamic and corticocortical connections of the
second somatic sensory area of the mouse. Journal of Comparative Neurology
265, 409–427.
Carvell, G.E., Simons, D.J., 1990. Biometric analyses of vibrissal tactile discrimination
in the rat. Journal of Neuroscience 10, 2638–2648.
Carvell, G.E., Simons, D.J., 1995. Task- and subject-related differences in sensorimotor
behavior during active touch. Somatosensory and Motor Research 12, 1–9.
Chmielowska, J., Carvell, G.E., Simons, D.J., 1989. Spatial organization of thalamocortical
and corticothalamic projection systems in the rat SmI barrel cortex.
Journal of Comparative Neurology 285, 325–338.
Clarke, W., Bowsher, D., 1962. Terminal distribution of primary afferent trigeminal
fibers in the rat. Experimental Neurology 6, 372–383.
Curtis, J.C., Kleinfeld, D., 2009. Phase-to-rate transformations encode touch in
cortical neurons of a scanning sensorimotor system. Nature Neuroscience
12, 492–501.
Descheˆnes, M., Timofeeva, E., Lavalle´e, P., Dufresne, C., 2005. The vibrissal system as
a model of thalamic operations. Progress in Brain Research 149, 31–40.
Devlin, J.T., Price, C.J., 2007. Perirhinal contributions to human visual perception.
Current Biology 17, 1484–1488.
Diamond, M.E., 2010. Texture sensation through the fingertips and the whiskers.
Current Opinion in Neurobiology 20, 319–327.
Diamond, M.E., von Heimendahl, M., Arabzadeh, E., 2008a. Whisker-mediated
texture discrimination. PLoS Biology 6, e220.
Diamond, M.E., von Heimendahl, M., Itskov, P., Arabzadeh, E., 2008b. Response to:
Ritt et al., ‘‘embodied information processing: vibrissa mechanics and texture
features shape micromotions in actively sensing rats’’ Neuron 57, 599–613
Neuron 60, 743–744; author reply 745–747.
Diamond, M.E., von Heimendahl, M., Knutsen, P.M., Kleinfeld, D., Ahissar, E., 2008c.
Where and what in the whisker sensorimotor system. Nature Reviews Neuroscience
9, 601–612.
Do¨rfl, J., 1985. The innervation of the mystacial region of the white mouse: a
topographical study. Journal of Anatomy 142, 173–184.
Ebara, S., Kumamoto, K., Matsuura, T., Mazurkiewicz, J.E., Rice, F.L., 2002. Similarities
and differences in the innervation of mystacial vibrissal follicle-sinus
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–40 39
complexes in the rat and cat: a confocal microscopic study. Journal of Comparative
Neurology 449, 103–119.
Fassihi, A., Akrami, A., Esmaeili, V., Diamond, M.E., 2012. Sensation of a – noisy
whisker vibration in rats: psychometric and neurometric analysis. Cosyne
Abstracts 2012, Salt Lake City USA.
Fortin, N.J., Wright, S.P., Eichenbaum, H., 2004. Recollection-like memory retrieval
in rats is dependent on the hippocampus. Nature 431, 188–191.
Gamzu, E., Ahissar, E., 2001. Importance of temporal cues for tactile spatialfrequency
discrimination. Journal of Neuroscience 21, 7416–7427.
Ganguly, K., Kleinfeld, D., 2004. Goal-directed whisking increases phase-locking
between vibrissa movement and electrical activity in primary sensory cortex in
rat. Proceedings of the National Academy of Sciences of the United States of
America 101, 12348–12353.
Gelbard-Sagiv, H., Mukamel, R., Harel, M., Malach, R., Fried, I., 2008. Internally
generated reactivation of single neurons in human hippocampus during free
recall. Science 322, 96–101.
Guic-Robles, E., Jenkins, W.M., Bravo, H., 1992. Vibrissal roughness discrimination is
barrelcortex-dependent. Behavioural Brain Research 48, 145–152.
Hafting, T., Fyhn, M., Molden, S., Moser, M.B., Moser, E.I., 2005. Microstructure of a
spatial map in the entorhinal cortex. Nature 436, 801–806.
Harris, J.A., Petersen, R.S., Diamond, M.E., 1999. Distribution of tactile learning and
its neural basis. Proceedings of the National Academy of Sciences of the United
States of America 96, 7587–7591.
Harvey, M., Bermejo, R., Zeigler, H., 2001. Discriminative whisking in the head-fixed
rat: optoelectronic monitoring during tactile detection and discrimination
tasks. Somatosensory and Motor Research 18, 211.
Hill, D.N., Bermejo, R., Zeigler, H.P., Kleinfeld, D., 2008. Biomechanics of the vibrissa
motor plant in rat: rhythmic whisking consists of triphasic neuromuscular
activity. Journal of Neuroscience 28, 3438–3455.
Hutson, K.A., Masterton, R.B., 1986. The sensory contribution of a single vibrissa’s
cortical barrel. Journal of Neurophysiology 56, 1196–1223.
Itskov, P.M., Vinnik, E., Diamond, M.E., 2011. Hippocampal representation of touchguided
behavior in rats: persistent and independent traces of stimulus and
reward location. PLoS ONE 6, e16462.
Jadhav, S.P., Wolfe, J., Feldman, D.E., 2009. Sparse temporal coding of elementary tactile
features during active whisker sensation. Nature Neuroscience 12, 792–800.
Jenks, R.A., Vaziri, A., Boloori, A.R., Stanley, G.B., 2010. Self-motion and the shaping
of sensory signals. Journal of Neurophysiology 103, 2195–2207.
Jones, L.A., Lederman, S.J., 2006. Human Hand Function. Oxford University Press, USA.
Jones, L.M., Lee, S., Trageser, J.C., Simons, D.J., Keller, A., 2004. Precise temporal responses
in whisker trigeminal neurons. Journal of Neurophysiology 92, 665–668.
Kepecs, A., Uchida, N., Zariwala, H.A., Mainen, Z.F., 2008. Neural correlates, computation
and behavioural impact of decision confidence. Nature 455, 227–231.
Kim, U., Ebner, F.F., 1999. Barrels and septa: separate circuits in rat barrels field
cortex. Journal of Comparative Neurology 408, 489–505.
Kleinfeld, D., Ahissar, E., Diamond, M.E., 2006. Active sensation: insights from the rodent
vibrissa sensorimotor system. Current Opinion in Neurobiology 16, 435–444.
Knutsen, P.M., Pietr, M., Ahissar, E., 2006. Haptic object localization in the vibrissal
system: behavior and performance. Journal of Neuroscience 26, 8451–8464.
Kouh, M., Poggio, T., 2008. A canonical neural circuit for cortical nonlinear operations.
Neural Computation 20, 1427–1451.
Kwegyir-Afful, E.E., Keller, A., 2004. Response properties of whisker-related neurons
in rat second somatosensory cortex. Journal of Neurophysiology 92, 2083–2092.
Lak, A., Arabzadeh, E., Diamond, M.E., 2008. Enhanced response of neurons in rat
somatosensory cortex to stimuli containing temporal noise. Cerebral Cortex 18,
1085–1093.
Lak, A., Arabzadeh, E., Harris, J., Diamond, M., 2010. Correlated physiological and
perceptual effects of noise in a tactile stimulus. Proceedings of the National
Academy of Sciences of the United States of America 17, 7981–7986.
Lavan, D., McDonald, J.S., Westbrook, R.F., Arabzadeh, E., 2011. Behavioural correlate
of choice confidence in a discrete trial paradigm. PLoS ONE 6, e26863.
Lederman, S.J., Klatzky, R.L., 1987. Hand movements: a window into haptic object
recognition. Cognitive Psychology 19, 342–368.
Lichtenstein, S.H., Carvell, G.E., Simons, D.J., 1990. Responses of rat trigeminal
ganglion neurons to movements of vibrissae in different directions. Somatosensory
and Motor Research 7, 47–65.
Lottem, E., Azouz, R., 2008. Dynamic translation of surface coarseness into whisker
vibrations. Journal of Neurophysiology 100, 2852.
Lottem, E., Azouz, R., 2011. A unifying framework underlying mechanotransduction
in the somatosensory system. Journal of Neuroscience 31, 8520–8532.
Maravall, M., Petersen, R.S., Fairhall, A.L., Arabzadeh, E., Diamond, M.E., 2007. Shifts
in coding properties and maintenance of information transmission during
adaptation in barrel cortex. PLoS Biology 5, e19.
Merleau-Ponty, M., 1945. Phe´nome´nologie de la perception. Gallimard, Paris.
Mitchinson, B., Arabzadeh, E., Diamond, M.E., Prescott, T.J., 2008. Spike-timing in
primary sensory neurons: a model of somatosensory transduction in the rat.
Biological Cybernetics 98, 185–194.
Mitchinson, B., Gurney, K.N., Redgrave, P., Melhuish, C., Pipe, A.G., Pearson, M.,
Gilhespy, I., Prescott, T.J., 2004. Empirically inspired simulated electro-mechanical
model of the rat mystacial follicle-sinus complex. Proceedings of the Royal
Society B: Biological Sciences 271, 2509–2516.
Morita, T., Kang, H., Wolfe, J., Jadhav, S.P., Feldman, D.E., 2011. Psychometric curve
and behavioral strategies for whisker-based texture discrimination in rats. PLoS
ONE 6, e20437.
Munk, H., 1881. Ueber die Functionen der Grosshirnrinde; gesammelte Mittheilungen
aus den Jahren 1877–80. Hirschwald, Berlin.
Murphy, R.A., Mondragon, E., Murphy, V.A., 2008. Rule learning by rats. Science 319,
1849–1851.
Neff, W.D., Diamond, I.T., 1958. The neural basis of auditory discrimination. In:
Harlow, H.F., Woolsey, C.N. (Eds.), Biological and Biochemical Bases of Behavior.
University of Wisconsin Press, Madison, pp. 101–126.
Nicholls, J.G., Martin, A.R., Fuchs, P.A., Brown, D.A., Diamond, M.E., Weisblat, D.A.,
2012. From Neuron to Brain. Sinauer, Sunderland, MA.
O’Keefe, J., 1976. Place units in the hippocampus of the freely moving rat. Experimental
Neurology 51, 78–109.
Peretz, I., 2006. The nature ofmusicfroma biological perspective. Cognition100,1–32.
Perkon, I., Kosir, A., Itskov, P.M., Tasic, J.F., Diamond, M.E., 2011. Unsupervised
quantification of whisking and head movement in freely moving rodents.
Journal of Neurophysiology 105, 1950–1962.
Petersen, R.S., Panzeri, S., Maravall, M., 2009. Neural coding and contextual influences
in the whisker system. Biological Cybernetics 100, 427–446.
Pons, T.P., Garraghty, P.E., Mishkin, M., 1992. Serial and parallel processing of tactual
information in somatosensory cortex of rhesus monkeys. Journal of Neurophysiology
68, 518–527.
Prescott, T.J., Diamond, M.E., Wing, A.M., 2011. Active touch sensing: an introduction
to the theme issue. Philosophical Transactions of the Royal Society of
London, Series B: Biological Sciences 366, 2989–2995.
Rhodes, D.H., Richmond, M.E., 1985. Influence of soil texture, moisture and temperature
on nest-site selection and burrowing by the pine vole, Microtus pinetorum.
American Midland Naturalist 102–108.
Rice, F.L., Mance, A., Munger, B.L., 1986. A comparative light microscopic analysis of
the sensory innervation of the mystacial pad. I. Innervation of vibrissal folliclesinus
complexes. Journal of Comparative Neurology 252, 154–174.
Romo, R., de Lafuente, V., 2012. Conversion of sensory signals into perceptual
decisions. Progress in Neurobiology, http://dx.doi.org/10.1016/j.pneurobio.
2012.03.007 in press.
Shoykhet, M., Doherty, D., Simons, D.J., 2000. Coding of deflection velocity and
amplitude by whisker primary afferent neurons: implications for higher level
processing. Somatosensory and Motor Research 17, 171–180.
Stuttgen, M.C., Schwarz, C., 2008. Psychophysical and neurometric detection performance
under stimulus uncertainty. Nature Neuroscience 11, 1091–1099.
Stu¨ttgen, M.C., Schwarz, C., 2010. Integration of vibrotactile signals for whisker-related
perception in rats Is governed by short time constants: comparison of neurometric
and psychometric detection performance. Journal of Neuroscience 30, 2060–2069.
Swanson, L.W., 1981. A direct projection from Ammon’s horn to prefrontal cortex in
the rat. Brain Research 217, 150–154.
Tafazoli, S., Di Filippo, A., Zoccolan, D., 2012. Transformation-tolerant object recognition
in rats revealed by visual priming. Journal of Neuroscience 32, 21–34.
Thorpe, S.J., Fize, D., Marlot, C., 1996. Speed of processing in the human visual
system. Nature 381, 520–522.
Torvik, A., 1956. Afferent connections to the sensory trigeminal nuclei, the nucleus
of the solitary tract and adjacent structures; an experimental study in the rat.
Journal of Comparative Neurology 106, 51–141.
Veinante, P., Deschenes, M., 2003. Single-cell study of motor cortex projections to
the barrel field in rats. Journal of Comparative Neurology 464, 98–103.
Vincent, S., 1912. The Functions of the Vibrissae in the Behavior of the White Rat.
University of Chicago.
Voges, D., Carl, K., Klauer, G.J., Uhlig, R., Schilling, C., Behn, C., Witte, H., 2012.
Structural characterisation of the whisker system of the rat. IEEE Sensors
Journal 12, 332–339.
von Heimendahl, M., Itskov, P.M., Arabzadeh, E., Diamond, M.E., 2007. Neuronal
activity in rat barrel cortex underlying texture discrimination. PLoS Biology 5,
e305.
Welker, C., Woolsey, T.A., 1974. Structure of layer IV in the somatosensory neocortex
of the rat: description and comparison with the mouse. Journal of Comparative
Neurology 158, 437–453.
Whitfield, I.C., 1979. The object of the sensory cortex. Brain, Behavior and Evolution
16, 129–154.
Williams, C.M., Kramer, E.M., 2010. The advantages of a tapered whisker. PLoS ONE
5, e8806.
Wolfe, J., Hill, D.N., Pahlavan, S., Drew, P.J., Kleinfeld, D., Feldman, D.E., 2008. Texture
coding in the rat whisker system: slip-stick versus differential resonance. PloS
Biology 6, e215.
Woolsey, T.A., van der Loos, H., 1970. The structural organization of layer IV in the
somatosensory region (SI) of mouse cerebral cortex. The description of a
cortical field composed of discrete cytoarchitectonic units. Brain Research
17, 205–242.
Wright, A.K., Ramanathan, S., Arbuthnott, G.W., 2001. Identification of the
source of the bilateral projection system from cortex to somatosensory
neostriatum and an exploration of its physiological actions. Neuroscience
103, 87–96.
Yamane, Y., Carlson, E.T., Bowman, K.C., Wang, Z., Connor, C.E., 2008. A neural code
for three-dimensional object shape in macaque inferotemporal cortex. Nature
Neuroscience 11, 1352–1360.
Yu, C., Derdikman, D., Haidarliu, S., Ahissar, E., 2006. Parallel thalamic pathways for
whisking and touch signals in the rat. PLoS Biology 4, e124.
Zoccolan, D., Oertelt, N., DiCarlo, J.J., Cox, D.D., 2009. A rodent model for the study of
invariant visual object recognition. Proceedings of the National Academy of
Sciences of the United States of America 106, 8748–8753.
Zuo, Y., Perkon, I., Diamond, M.E., 2011. Whisking and whisker kinematics during a
texture classification task. Philosophical Transactions of the Royal Society of
London, Series B: Biological Sciences 366, 3058–3069.
M.E. Diamond, E. Arabzadeh / Progress in Neurobiology 103 (2013) 28–4040