OUP user menu

The sensory hand

Edward G. Jones
DOI: http://dx.doi.org/10.1093/brain/awl308 3413-3420 First published online: 28 November 2006

We explore the world around us with our eyes and hands but, as pointed out by John Napier in Hands (1980), only one of these permits us to see around corners and in the dark. The exploratory capacity of the hand brings to the tactile sense a quality that transcends all the other senses and led Bichat in the early days of the 19th century to refer to touch as the only active sense. An inert hand receives an impoverished sensory input and is but a poor transmitter of information about an object placed in it to the centres for perception. But to observe a skilled Braille reader translating series of raised dot patterns into meaningful language at a rate of up to 100 words per min is to recognize not only the high resolution sensory capacities of the active human hand but also the capacity of the neural signals generated by the mechanoreceptors in the moving fingers to gain rapid access to the highest cognitive centres. This is on the input side. On the output side, apart from the chimpanzees who exhibit a limited capacity, humans are the only species that can communicate meaningfully with the hands. It is not without significance that in naming the individual fingers, medieval anatomists dubbed the middle finger impudicus or obscenus. (medieval lawyers, too, but that is another story).

The hand and the somatosensory system of which it is the handmaiden have never attracted the same extensive and long continued treatment as the eye and the visual sense. Vernon Mountcastle's book, The Sensory Hand, nevertheless, comes in a tradition that commenced with Sir Charles Bell's Bridgewater Treatise of 1832 (The Hand its Mechanism and Vital Endowments as Evincing Design), a pre-Darwinian perspective on the place of the forelimb appendage in the broader economy of nature. Bell like Napier and others after him, for example, Frederic Wood Jones in his influential The Principles of Anatomy as Seen in the Hand (1919), was principally focused on the hand as a motor organ, fascinated as they were by the almost unlimited capacity of the human hand for precision and dexterity of movement. In this, they saw the hand as an executive organ of the brain. To Napier, the hand was the mirror of the brain and ‘there can be no such combination as dextrous hands and clumsy brains’. For Wood Jones, ‘It is not the hand that is perfect, but the whole nervous mechanism by which the movements of the hand are evoked, coordinated and controlled’. All recognized nevertheless the important relationship between tactile sensation and the motion of the hand, Bell summing up one of his chapters with ‘… . how happily the hand is constructed: in which we perceive the sensibilities to changes of temperature, to touch, and to motion, united to a facility of motion in the joints, for unfolding and turning the fingers in every possible degree and direction, without abruptness or angularity, and in a manner inimitable by any artifice of joints and levers’.

None can doubt the almost unlimited manipulative capacities of the human hand, capacities endowed upon us by our early forebears who, in not having passed through an arboreal phase, did not have to contend when returning to the ground with a hand adapted to brachiation. This imposed upon the great apes not only the obligation of knuckle walking but also a reduced capacity for fine manipulative skill and tactile exploration. The elongated palm and the abbreviated thumb of the apes provides a far greater capacity for swinging from limb to limb than Tarzan could ever have had, and the ape's ability to lock the fingers into the palm in gripping a rope or vine ensures that Tarzan would never have won a tug of war with a chimp. But only Tarzan could hold a grape between the thumb and index finger tips.

Our forebears, like those of the Old World monkeys, kept their front paws firmly on the ground and, like the monkeys, we can still hold the palm of the hand down firmly on a flat surface with both the wrist and fingers fully extended—something the apes cannot do. The reduced thumb of the great apes, although as readily rotatable as that of a human, cannot have its tip brought into complete apposition with the tip of the index or other fingers. The precision grip is thus imperfect in comparison with that of the human in which the distal palmar pad of the thumb can be brought into perfect opposition to the distal pad of any other finger. But the apposition of the finger pads is not merely an instrument of manual dexterity. It is as much an implement for bringing two richly innervated sensory surfaces to bear upon an object of manipulation. From the movements of the opposed finger tips, the brain learns about the shape, size, softness or hardness, texture, slipperiness, stickiness and temperature of an object palpated. The papillary ridges of the opposed tips facilitate the precision grip but they are also instruments of tactile exploration, and sweat oozing from the openings of the sweat glands located on the ridges while facilitating grip also enhances the senses of touch and pain.

Questions of dexterity aside, we still know that the sensory capacities of the monkey and human hand are in most ways identical. This undoubtedly depends upon the virtually identical natures of the various kinds of sensory receptors located in the hand in monkeys and humans (and probably apes, too, but no one has explored this in any significant way). In The Sensory Hand, Vernon Mountcastle explores the relationships between the capacities of these receptors to signal different qualities of a haptic stimulus and of the cerebral cortex in converting the signals into elements of perception. His groundwork lies in the important series of experiments carried out in the 1970s and 80s in his own laboratory on the nerve fibres of the Old World monkey's hand and the impressive findings made on human nerve fibres by the technique of microneuronography introduced by Hagbarth and Vallbo, during the same period. When allied with observations made on the behaviour of nerve cells of the post-central gyrus and superior parietal lobule in awake, behaving monkeys, a methodology also pioneered by Mountcastle in the 1970s, the data as a whole permit Mountcastle to present to the reader an insightful account of how neural representations of peripheral stimuli are likely to be transformed into representations necessary for perception and behavioural responses. He is quick to point out that he has few answers to the larger issues of how cognition, self-awareness and consciousness itself emerge from these operations of the higher brain centres, but the strength of his approach is that it is firmly grounded in experimental observations and the writer is not one who is willing to venture much beyond what the neurons may be telling us into the realms of speculative neuro-mumbo-jumbo masquerading as ‘theory’.

The skin, muscles and joints of the hand are innervated by as many as 14 different types of nerve fibre (classifications vary as to the exact number), each transducing and encoding specifically different properties of stimuli reaching or obtained by the hand. Some of the fibres end in relation to specialized sensory end organs. Others, primarily the thin fibres, end freely. Together, they form an overlapping spatial mosaic of receptors with selective transducing functions, some mechanical, some chemical and/or thermal. Selectivity resides in the terminal nerve fibre itself, not in any other cellular elements associated with the terminations. The molecular mechanisms that determine the transducing properties of nociceptive afferents have recently become well known following the cloning of the vanilloid receptors for which heat is the natural ligand. Heat influences the probability of opening of a non-selective cation channel expressed mainly in the unmyelinated C fibres. Other non-vanilloid receptors expressed primarily in the thinly myelinated Aδ fibres are also non-specific cation channels activated by heat in the noxious range. Low threshold mechanical sensitivity mediated by thicker myelinated fibres is undoubtedly mediated by cation channels gated by tension of the cell membrane and akin to those identified in invertebrates. For these, associated connective tissue cells such as those in the Pacinian and Meissner corpuscles may direct the stretch in an orderly way towards the receptor channel, and presumptive neurosecretory cells in Merkel's tactile discs may lend a chemical modulator as well, but it is the fibre itself that is the effective receptor.

The innervation of the skin of the hand and fingers by a rich and varied constellation of mechanoreceptors underlies the remarkable capacities of the hand as an implement of active touch. Weber taught us in the 1830s that the texture of a surface can only be identified by a moving finger and recent research quoted by Mountcastle shows that we use a rather small set of stereotyped scanning movements in order to achieve this. In doing so, we can detect textures with element spacings as small as 0.1 mm and differences of <0.75 mm in gaps, gratings, orientations and elevations on the surface of an object manipulated by the fingers. The fingers and hand are capable of resolving differences of ∼10% in curved objects; differences of 4–5° in orientated objects; differences of 2–3 mm in stimuli moving in the same direction and differences of ∼14° in their orientations. Vibratory stimuli with amplitudes as small as 0.01 μm at 300 Hz can be detected and we can discriminate 2–3 Hz differences at base frequencies of 20–40 Hz. The capacity of the various kinds of mechanoreceptive afferents in the hand to provide the signals that the nervous system relies upon to make these discriminations are a dominating theme of The Sensory Hand.

Mountcastle, in drawing upon the enormous body of data collected by microelectrode recordings from single fibres in the nerves of the Old World monkey hand carried out by he and his colleagues, and the remarkable microneuronographic studies of individual fibres in human nerves carried out mainly by Swedish investigators, provides one of the best accounts currently available of the range and sensory capacities of each type. As Mountcastle emphasizes, with no more than one potential exception, the types and their specificities in encoding different qualities of a stimulus are identical in humans and monkeys and the sensory properties inferred from recording the specificities of response in single fibre types in monkeys and humans are amply confirmed by the sensory experiences evoked by stimulation of isolated fibres in the nerves of humans.

For the glabrous skin of the palm and palmar surfaces of the fingers, there are four kinds of thicker, myelinated Aβ fibres conducting in the range of 25–75 m/s and ending specifically in relation to the specialized non-neural cells of Meissner corpuscles, Pacinian corpuscles, Merkel cell–neurite complexes or Ruffini corpuscles, the last mentioned perhaps peculiar to the human hand. These fibres are all sensitive at low threshold to non-noxious mechanical stimulation of the skin and grade their frequencies of discharge to the intensity of a stimulus. Those associated with Meissner or Pacinian corpuscles are rapidly adapting and discharge primarily at onset or offset of a stimulus. They are therefore especially qualified to signal stimulus transients such as contact, movement, velocity and vibration. Stimulation of identified Meissner afferents in humans evokes sensations of moving stimuli and low frequency vibration. Stimulation of Pacinian afferents evokes sensations of contact, movement and high frequency vibration. Fibres associated with Merkel cell–neurite complexes or Ruffini corpuscles are slowly adapting and discharge at stimulus onset and (usually at a lower rate) throughout the application of a stimulus. They are therefore qualified to signal stimulus continuity such as pressure, displacement and skin stretch. Stimulation of Merkel cell-associated afferents in humans evokes sensations of contact, pressure, texture and two-dimensional form. Stimulation of Ruffini afferents, somewhat surprisingly, evokes no obvious sensations. Thinly myelinated Aδ fibres (5–40 m/s) and unmyelinated C fibres (0.5–2 m/s) form at least six receptor classes specific for cooling, warming, noxious heat or cold, destructive mechanical or mixed noxious stimuli. All are relatively insensitive to the kinds of innocuous mechanical stimuli transduced at low threshold by the Aβ fibres and do not reach their mechanical thresholds until the Aβ fibres are firing at maximal rates. Stimulation of the isolated fibres in humans evokes sensations of coolness, warmth, intense mechanical pain, cold pain, heat pain or slow burning pain, but with an emphasis on pricking pain at short latency for the Aδ fibres and on slow, burning but accurately localized pain for the C fibres. All of this betokens a high degree of receptor specificity of the kind envisaged by earlier generations of sensory physiologists and psychophysicists but which has needed years of experimental work and human microneuronography to reach maturity.

Mountcastle's hand is essentially a glabrous one and he is at his best in describing the properties of the low threshold mechanoreceptors in the glabrous skin and how these properties in many instances provide a quasi-isomorphic representation of stimuli for presentation to the first somatic sensory area of the post-central gyrus. The discharges of the rapidly adapting fibres innervating Meissner corpuscles can become phase locked to cyclic stimuli over the range of 2–50 Hz and when stimulated in humans elicit a sense of flutter. Pacinian afferents can entrain stimuli at 50–500 Hz and when stimulated elicit a sense of vibration. Each of these afferents has an absolute threshold for detection of a stimulus and a tuning threshold for phase locking to a cyclical stimulus. Below these thresholds, a cyclical stimulus is reported to elicit in humans a sensation of local pressure only. The slowly adapting fibres innervating Merkel cell complexes signal by the spatial pattern of their afferent discharges the form and texture of objects palpated by the hand or fingers. In their case the population response of all the fibres activated in an area of skin contact with an object gives a remarkably isomorphic representation of the contours of the object, best seen, for example, when a series of dot patterns that form letters of the alphabet are moved over a monkey's fingers. The resolving power of the slowly adapting population response is close to that predicted from the innervation density of the slowly adapting afferents in the glabrous skin. This response characteristic of the slowly adapting afferents, when allied with movement of the fingers, is perfectly adapted to the requirements for reading Braille patterns in humans.

Despite the remarkable stimulus-related specificity of the fibres innervating the skin of the hand, Mountcastle is quick to point out that the thresholds for activation of all the various afferents, low threshold and high threshold included, overlap. It is, thus, unlikely that in the normal manipulation of an object or even in the reception of a passively applied stimulus that an individual receptor class is activated in isolation. Even in the absence of a thermal or noxious content in a stimulus, the higher threshold, fine fibred receptors should still work in a manner coordinated with the lower threshold, thick fibred mechanoreceptors to signal the full content of a stimulus acquired by the hand. Caressing a loved one with the hand is accompanied by more than just a sensation of touch. Or at a more banal level, an object weighed by the hand feels heavier when warm than when cold.

Mountcastle's sensory hand is also largely a cutaneous one. His attention is focused mainly on the receptors located in the epidermis, dermis and immediate subcutaneous environment, drawing, as mentioned above, on his enormous fund of personal knowledge to build up a picture of their individual and collective roles as agents of somatic sensation. The deep receptors of the muscles, joints and aponeurotic membranes of the hand receive but passing mention. Here, one might have expected more. It is hard to believe, for example, that the tiny lumbrical muscles, attached from the deep digital flexor tendons to the terminal expansions of the extensor tendons, can have much to do with initiating movements of the fingers. But when one considers that they, along with the interossei, are packed with those exquisitely sensitive stretch receptors and monitors of muscle length, the muscle spindles, it is difficult to escape the conclusion that they may be capable of signalling finger position. The role of the primary afferents of the muscle spindles in the conscious perception of joint position and movement is something that has gone in and out of favour over the years. Following the clever work of Matthews and colleagues in the 1970s, in which they showed that artificial stimulation of the primary afferents could result in clear-cut illusions of movement and joint rotation, the minimal kinaesthetic deficits that result from anaesthesia or artificial replacements of human joints, and the demonstration of a direct lemniscal pathway from the primary afferents to area 3a of the somatosensory cortex, the argument seemed to have been settled in favour of the spindle afferents as mediators of conscious proprioception. For the fingers, however, it may be more complicated and the sense of finger position may as much depend on cutaneous receptors in the hand and on the backs of the finger joints as upon those in the bellies of the long muscles of the forearm that move the fingers. Recent work has shown, for example, that receptors in the skin overlying the dorsal aspects of the finger joints are especially sensitive to the stretching of the skin that occurs during joint movement. And there is strong activation of all four classes of cutaneous mechanoreceptor afferents during voluntary isotonic finger movements. Microneuronographic studies reveal that joint afferents by themselves possess a very limited capacity to signal joint position and movement but, on the other hand, they also reveal that specific stimulation of muscle spindle primary afferents is not accompanied by any perceived sensation of movement.

The emphasis on the low threshold cutaneous mechanoreceptors in The Sensory Hand is not surprising given that so much of the fundamental work on these receptors and their central projections was carried out over a lifetime of study by Vernon Mountcastle and continued on in that of his colleagues, notably the late Dr Kenneth Johnson, to whom the book is dedicated. But Mountcastle also exhibits a sure touch in navigating his way through the fine fibred high threshold systems in which the hard data recorded by workers such as Edward Perl and William Willis and their colleagues have in recent years been overlaid with much speculative thinking and not a little charlatanry. Mountcastle's descriptions of the six or more high threshold receptors, the central terminations of the fibres in the dorsal horn and the neural operations of the dorsal horn itself are ineluctable in their clarity and should be instructive to anyone who wishes to obtain an objective overview of these systems. He is particularly good on the properties of the high threshold afferents themselves and on the local changes that follow damage to the skin and which can lead to the phenomena of sensitization, hyperalgesia and allodynia, along with the underlying bases in the release of activating molecules from damaged nerve terminals. If this is textbook stuff, it is not often seen nowadays and rarely has it been presented so lucidly.

The measured approach to the pain system continues as Mountcastle follows the somatosensory pathways through the spinal cord, brainstem and thalamus to the cerebral cortex. Scholarly rectitude sometimes obliges him to refer to unrepeatable anatomical work on the ascending fibres and on the pain regions of the posterior thalamus, visualized by those with a deep faith in morphological apparitions vouchsafed only to themselves. They have presented a case for the relay of pain through a very restricted part of the thalamus to areas of cerebral cortex far removed from the primary somatosensory area. Viewing all the available physiological, anatomical and functional imaging data which proponents of the new Jerusalem have failed to do, Mountcastle arrives at the reasonable conclusion that stimulus amplitude for pain is relayed through the ventral posterior nucleus of the thalamus to the post-central gyrus, while pain intensity may be more a function of distributed areas of the lateral sulcus and anterior cingulate regions with inputs from equally distributed regions of the thalamus, and the affective overtones a function of more posterior cingulate regions also with inputs from some of the multiple regions of the thalamus in which nociceptive specific fibres arising from the marginal layer of the dorsal horn terminate.

On taking his reader into the central nervous system, Mountcastle, like Charles Bell before him, begins to lose sight of the hand and adopts a far broader perspective on the somatosensory pathways and of haptic sensation in general. He is impressed, as anyone familiar with the territory is, that despite enormous divergence and convergence both within and across the lemniscal and spinothalamic pathways, modality and place specificity remain intact. The properties of the receptor-specific classes of peripheral afferents are preserved in the responses of neurons in the spinal cord, dorsal column nuclei, thalamus, and for some even in the primary somatosensory cortex itself. There is little loss of specificity and little decrement in response sensitivity. The divergence provides one of the bases of the profound capacity that the system has for plasticity in the face of even extreme degrees of peripheral or central deafferentation. It is possible for example to destroy >15% of the representation of the hand in the ventral posterior thalamus before any contraction is evident in the extent of the hand representation in the post-central gyrus. Mountcastle reviews all the pertinent literature on this subject in a specific chapter of his book. The divergence and convergence that becomes revealed in receptive field shifts and shifts in representational maps when the system is perturbed has to be, according to Mountcastle, subliminal when the properties of relay cells throughout the pathways are examined under the highly restricted controlled conditions of the physiological experiment and in which the parallel, receptor-specific channels are revealed. He is undoubtedly correct here for the extracellular recordings of a typical experiment rarely if ever reveal the subthreshold inputs from wider areas of skin and perhaps even from different receptors that become evident as post-synaptic potentials in intracellular recordings. We are still waiting to find out what contribution such subthreshold inputs make to the sensory performance of an animal or person whose hand is engaged in a complex tactile manipulation. Mountcastle thinks that they will all contribute in some manner to the overall haptic percepts accompanying this manipulation.

As to mechanisms supporting receptive field- and receptor-specificity through a highly divergent system, we are still pretty much in the dark about mechanisms that prevent much supraliminal convergence across the parallel channels ascending from each of the low threshold and from the high threshold receptors at levels below the cortex. Apart from the usual candidates of synaptic specificity, afferent inhibition, cortical feedback and attention, Mountcastle likes the division of the sensory relay nuclei of the thalamus identified by the author of this review into two compartments: a core compartment in which the fibres of the lemniscal pathway and of its equivalents in the visual and auditory system terminate in a high degree of topographic and modality specific order on chemically distinct relay cells; and a spatially discrete matrix compartment in which those of the spinothalamic system and its equivalents terminate more diffusely on relay cells with a different chemical identity. The projection of the relay cells in the two compartments to different layers of the cerebral cortex also has implications for the way in which ascending information in the lemniscal and spinothalamic pathways is handled at the cortical level. Although an absence of convergence at the synaptic level across compartments has not been ruled out, the compartment organization tends to keep the two ascending systems separate and Mountcastle would like to see the core and matrix organization extended to the lower relay stations of the somatosensory pathways.

On arriving at the cortex, Mountcastle returns to the hand and attempts to analyse in depth the manner in which the highly receptor-specific inputs are converted into a code recognizable beyond the arrival platform for these inputs in the post-central gyrus, how the output is channeled through a highly distributed network of intercortical connections, and how the sensory input from the hand is integrated with cortical mechanisms underlying visually guided behaviour and manual dexterity. Here, he is again at his best for the first and last of these are areas in which his own research has been fundamental. All modern studies of parietal lobe function in awake behaving monkeys stem from his seminal work of the 1970s and 80s.

Throughout the lemniscal pathway from dorsal column nuclei through ventral posterior thalamus to the post-central gyrus, the three major low threshold mechanoreceptive systems remain essentially separate and form parallel channels that may have privileged access to the centres for perception. In areas 3b and 1 of the post-central gyrus many neurons have stimulus–response properties that virtually replicate those of the peripheral afferents. Approximately 75% of the neurons encountered in single unit recordings in monkeys reveal a dominating input from rapidly adapting peripheral mechanoreceptors of the Meissner kind, about 14% have a dominating input from the Merkel cell-innervating type of slowly adapting afferent and about 6–8% have dominating inputs from Pacinian afferents. For the cortex, slowly adapting Merkel cell-innervating afferents provide graded signals about the degree of skin indentation, i.e. pressure. They are especially sensitive to the movement of stimuli across the skin and the discharges of a population of such afferents engaged by a moving stimulus signals the spatial form and surface structure of objects grasped by the hand or applied to the finger tips. Rapidly adapting Meissner afferents signal with precision the minute alternating skin movements perceived as the sensation of flutter. Their sensitivity to slippage of a stimulus makes them good candidates for feedback sensors regulating the control of grip. The rapidly adapting Pacinian receptors resolve spatial detail poorly but are exquisitely sensitive to high frequency sinusoidal stimuli sensed either directly or through a hand-held tool. No neurons with properties suggestive of inputs from slowly adapting Ruffini-type receptors have yet been recorded in the post-central gyrus of the monkey.

What are the critical transforms that the post-central gyrus imposes on the receptor-specific classes of input that it receives? The responses of single slowly adapting neurons in the middle, thalamically innervated, layers of these areas are as strongly isomorphic as the population response of the Merkel cell-innervating type of slowly adapting afferent when letters of the alphabet are scanned across the finger pads. But Mountcastle, who has always had an unerring feel for the dynamics in the operations of the nervous system, recognizes that it must be the onset response of the slowly adapting afferents that is the critical signal for perception. A skilled Braille reader, for example, can detect individual patterns in <100 ms, a period over which the afferents are still discharging in a sustained manner. For cortical cells receiving rapidly adapting inputs, the neurons respond to sinusoidal stimuli with phase-locked discharges, the action potentials being clustered around a portion of the stimulus wave and their density waxing and waning over the course of the sinusoid. In them, the abrupt tuning thresholds of the primary afferents are lost and there are no significant changes in the rates of impulse discharge in relation to two stimuli that can be perceptually discriminated with ease. Hence, for these neurons, a change in the periodicity of response is the likely neural code for the discrimination of frequencies in the sense of flutter. Cells in the middle layers of the cortex display cyclically entrained activity in response to stimuli applied to the glabrous skin of the hand at 100–300 Hz and thus activate Pacinian afferents, suggesting that for the sense of vibration the code may be one of periodic entrainment. This has been less well studied, however. As with the peripheral receptors, complex features of a stimulus that are easily perceived, such as the orientation, shape and movement (including its direction) of a stimulus applied to or acquired by the hand are less well coded in the responses of single cortical neurons. Individual neurons, especially in the supra- and infragranular layers, may exhibit some specificity for these features but it is again the responses of a population of cortical cells, probably located in multiple cortical columns and probably in more than a single area of the post-central gyrus, which provide a basis for these discriminations.

Mountcastle wants to know how the receptor specificity (he prefers to call it modality specificity) that is preserved from the receptor all the way to its ‘exquisite representation’ in the post-central gyrus is preserved in its transmission to an ‘equally precise component of perceptual experience’. To study this, he is obliged to turn to the connections emerging from the various fields of the post-central gyrus and here we can only admire his ability to make some degree of sense of an incomplete and confusing literature. Few of us have ever been able to cast the intercortical connectivity of the somatosensory areas in the same hierarchical light as that which pervades studies of the visual areas. The post-central areas are connected across the central sulcus with the motor and pre-motor areas and both departing and returning connections terminate in middle layers of the cortex. Since neither projects only to layer I, the hallmark according to our visual friends of a ‘feedback’ connection, both must be construed as ‘feedforward’ connections. Other connections pass unidirectionally out into the association areas of the superior parietal lobule in which the somatosensory input gains access to areas concerned with the direction of gaze and the projection of the arm into extrapersonal space. Others, some uni- and others bidirectional, are directed to various fields including the second somatosensory area in the Sylvian fissure and adjacent to the insula. Depending on whose anatomy one is willing to believe, these ultimately channel somatosensory information into the amygdala and limbic system. Mountcastle, too, cannot make much hierarchical sense of the connections and casts them, interestingly, in the light of a distributed and highly re-entrant system in which there are ‘nodes’ likely to be preferentially accessed by parallel pathways within the system of highly distributed connections. In these nodes, certain tasks may be performed, different perceptual experiences determined and behaviours produced. For Mountcastle, somatosensory processing in the cerebral cortex may be unconstrained by the strict hierarchical structure that researchers have tended to impose upon the visual system. It may be sometimes hierarchical, sometimes parallel and sometimes distributed but always should involve interactions between the cortex and underlying thalamus—something visual physiologists working in cortical areas beyond the striate area rarely contemplate these days.

Mountcastle is persuaded by experiments in Old World monkeys that suggest that the somatosensory areas of the lateral sulcus depend upon input from the post-central gyrus and are silent without it. This leads him to believe that serial processing from the first somatosensory area is a feature of this set of connections. Although recognizing that experiments in New World monkeys show that the areas of the lateral sulcus can indeed be activated just as effectively as those of the post-central areas by inputs from the ventral posterior thalamus, he is prepared to write this off as a less evolved feature of the primate brain, indicative of parallel rather than serial processing and perhaps shared by the New World monkeys with the prosimians. I am less convinced. Despite their separation by some 30 or 40 million years from their Old World relatives, there is nothing I have seen in the brains of New World monkeys to suggest that at this rather crude level of organization, there is anything substantially different about how the thalamus and cortex are interrelated. I would prefer to see the original experiments in macaques, upon which Mountcastle places such credence, repeated.

Some degree of hierarchical processing can be detected in the fields of the post-central gyrus and in the anterior parietal cortex of area 5 immediately behind them. Older lesion studies coupled with newer functional imaging studies indicate that area 3b is concerned with tactile detection, area 1 with texture discrimination, area 2 with spatial contour recognition and area 5 with the processing of the form of objects held in the hand. Lesions of area 3b are associated with more global deficiencies in somatic sensory sensation than lesions of the other areas and Mountcastle sees this as indicative of area 3b being the major source of inputs to areas 1, 2 and 5. I am less certain; I do not believe that there is sufficient evidence to rule out the thalamus as a contributor as well. Mountcastle would like to rule out area 3a as a mediator of joint and position sense. Given its remarkably specific input from muscle spindle primary afferents, I am uncertain about this too.

There is convergence as we pass out of the first somatic sensory area into area 5. All hand-related neurons in areas 3a, 3b, 1 and 2 of the first somatosensory area have strictly contralateral receptive fields that are small, the cutaneous ones restricted to tiny areas on a single finger or to a localized portion of the palm. Responses of neurons in areas 3b and 1 replicate those of the primary afferent fibres innervating Meissner, Merkel and Pacinian end formations. Area 5, to which these fields project, acquires neurons with bilateral and multi-finger receptive fields and selectively sensitive to the three-dimensional form of the (usually) contralateral hand and fingers. They are also strongly influenced by attention and the intent to move. It is here that we begin to see the linkages between the sensory and motor systems that make the hand an effective instrument of active touch.

Beyond area 5 we start to see the involvement of the guiding visual system as well. For the grasping and manipulating hand, the more posterior parietal areas, especially those in the banks of the intraparietal sulcus of monkeys, are key areas where visual and somatosensory afferent signals come together to define not only the location but also the salience of an object in extrapersonal space. Here, the programmes for the generation of an appropriate arm posture for reaching and a hand posture for grasping an object are generated and transferred to the pre-motor and motor areas of the frontal lobe. Current evidence points to the existence of neurons and perhaps whole areas of the intraparietal sulcus being selectively concerned with either reaching or with grasping. Grasping commences during the phase of projection of the arm into space and the posture of the hand becomes adapted to the visually determined or remembered shape of the object to be grasped; the grasping force, the precision of the grip and the lifting force applied when the hand reaches its target depend in the first instance on the visual interpretation of the object, presumably commenced in the posterior parietal cortex. But the force ratio just sufficient for lifting and capable of being rapidly adapted to surface texture and slippage is under exquisite control by virtue of the continuous afferent signals being fed to segmental and suprasegmental levels of the nervous system by the low threshold mechanoreceptor afferents in the palm and fingers. One suprasegmental level is likely to be direct connections from the post-central to the pre-central areas.

The puzzle for scientists is to determine how in the parietal cortex the neural signals for location of a visually identified object, formed as we know in eye-centred coordinates, are transformed into a set of neural signals representing arm-in-space and hand-object coordinates. Mountcastle, the scientist who initiated studies of the parietal cortex in awake behaving monkeys in which these issues first became capable of being investigated electrophysiologically, has a good deal to say about them and what he says can be read with profit. Although, as expected, contralateral neglect, inattention, lack of exploratory movements and deficient grasping movements appear in monkeys with parietal lobe lesions—a real but ‘pallid representation of the human’ parietal lobe syndrome, according to Mountcastle—the sensations of elementary touch and finger motion are clearly preserved. In this, we have to seek other pathways that permit the areas of the post-central gyrus, to which the low threshold mechanoreceptors of the hand project, to gain privileged access to the centres for perception.

Throughout this remarkable book, Vernon Mountcastle is primarily concerned with the contributions that the experimentally definable properties of peripheral receptors and the central neurons with which they are connected make to sensory perception and its behavioural sequelae. For him, as for all experimentalists, it is axiomatic that the selectivity of the receptors and the neural codes of amplitude, rate and spatio-temporal order that they transmit to the cerebral cortex are fundamental to any attempt to understand the nature of sensory perception. Not for him the position adopted by those who, if their views can be understood at all, fail to see how the brain uses the responses of nerve cells to support behaviour. A recent review by Alva Noë in the Times Literary Supplement (May 5, 2006) of Brain and Visual Perception, the work of another pair of distinguished sensory physiologists in the Mountcastle mould, David Hubel and Torsten Wiesel, concludes that the capacity of retinal ganglion cells and/or the neurons to which they connect centrally to break down the elements of a complex visual scene into signals recorded experimentally as centre-surround receptive fields, sensitivity to orientated bars of light and dark or to drifting gratings going in and out of phase can have little or nothing to do with the ‘brain-basis of vision’. Did Nature evolve these exquisite mechanisms for rendering the features of the external world into elements capable of being interpreted by a brain that faces it only through the medium of action potentials, solely for the occupation of experimentalists? All this would be bad enough if it were not accompanied by coyly fastidious little shudders at the thought that Hubel's and Wiesel's experiments involved retracting the eyelids of an animal, or, heaven forbid, suturing the eyelids closed! What frisson would Dr Noë experience if he were to learn that some of Vernon Mountcastle's most important work was carried out in monkeys whose heads had been deafferented?

Mountcastle recognizes, as we all should, that the recording of the stimulus–response properties of single neurons in the sensory and motor systems, a form of experimental endeavour that has characterized so much of modern neurophysiology, gives a predominantly static view of the pathways to perception. It is also one that does not readily permit the integrated activities of populations of single neurons, each with its own specially coded message about certain components of a complex stimulus, to be understood, except by making inferences about the population response from observations of the single units. The static properties are those of place, that is, the receptive field on the skin or in extrapersonal space subtended by a neuron's ramifications or by the collection of neurons feeding it, and of modality, that is, for the somatosensory system the receptor specificity that Mountcastle elaborates on with such skill in his book. Place and modality are properties that are hard-wired into the system and only modifiable if at all under extreme, usually pathological, conditions. In order to deal with the world around it, the brain has to take into account the dynamic properties of the inputs that it receives over these labelled lines. Dynamic properties are those that permit neurons in the sensory systems to transmit information about the amplitude of a stimulus in the intensity of discharge, about its temporal characteristics in the temporal order of discharges and about its spatial characteristics by the spatial and temporal pattern of discharges in populations of neurons. Intensity coding, temporal coding and population coding, allied with the static properties of place and receptor specificity, are the means whereby the exploring hand signals the details of what it encounters to the cerebral cortex. After that, it is still largely guesswork as to how these signals are converted to elements of conscious sensory experience, guesswork that still permits some to view cognition and consciousness as non-biological phenomena. Although the dynamics of neuronal performance in sensory processing can sometimes be predicted by reconstructing population responses from the recorded activities of single neurons, as Mountcastle repeatedly points out, only the simultaneous recording of hundreds, perhaps thousands of neurons in awake monkeys performing behaviourally meaningful tasks, perhaps allied with functional imaging of much higher resolution than currently available, can lead us to a better understanding of how activation of peripheral neurons exposed to the complex world around us is translated into perceptual experience. This is the challenge for the current generation of sensory physiologists.

Vernon Mountcastle tells his story with erudition and authority. For those who know him, this is a text in which they will be able to hear the man speak. Harvard University Press has done a little better by its author than in his previous Perceptual Neuroscience, the Cerebral Cortex (1998) in which the reproduction of the figures was disgraceful. But the editing of the present volume has been equally poor: many very unfortunate line breaks, inappropriate spellings that pervade the text and probably stem from early autocorrections, misalignments of tables, not a single accent on any of the foreign names, and an index that is rudimentary at best. It seems doubtful that the text of this book was touched by human hands after leaving the capable ones of Vernon Mountcastle. One of the most important figures in modern neurophysiology deserves better than this.