The Feedforward Model Of Touch

Like its counterpart model of vision, the feedforward model of touch has its origins in research conducted during a period that led to fundamental discoveries in sensory physiology.78-79 At the time of the model's conception, sensory physiology was swept by a revolutionary experimental paradigm: the single neuron recording. By taking advantage of the ability to record extracellular activity of a single neuron, in both anesthetized and awake animals, sensory neurophysiologists began to build what continues to be the most accepted theory of how the mammalian somatosensory system operates. What follows is a brief account of this model. For a more thorough account, see the review by Dykes.29 As expected, the main assumptions of this model clearly reflect the main neurophysiological approach used at the time the model was first conceived. Thus, in most accounts of the FF model of touch, the functional unit of computation is the single neuron, which is invariably described as a feature extractor, i.e., an element that is narrowly tuned to respond to a single feature (or a restricted set of features) of a tactile stimulus.

The central argument of the FF model of touch revolves around the well-known fact that the somatosensory system contains parallel, feedforward pathways that connect peripheral tactile receptors located throughout the body to the neocortex. These pathways carry information generated from a broad range of peripheral receptors, which are responsible for transducing different types of energies (mechanical, thermal, and chemical) that impact the body surface or are generated within the body (by proprioreceptors) into trains of action potentials. According to the FF model of touch, the role of these specialized peripheral receptors is to decompose complex tactile stimuli into their primary features. This feature decomposition is carried out by the differential tuning properties of a large variety of specialized mechanical, thermal, nociceptive, polimodal, and deep receptors (also known as proprioceptors) that send their outputs to the CNS through parallel ascending soma-tosensory pathways. Following a series of synapses in highly segregated subregions of the spinal cord, brainstem, and thalamus, these parallel streams of information terminate in the primary somatosensory cortex.62

The most extreme version of the FF model purports that each group of nerve fibers carrying the output of a particular class of somatosensory receptors defines an independent, feedforward labeled-line that faithfully conveys specific tactile information all the way to the neocortex.29-122 Thus, in addition to the well-accepted segregation between the dorsal-column/medial lemniscal system, which is specialized in conveying information from low threshold mechanoreceptors and proprio-ceptors, and the spinothalamic tract, which carries information from high-threshold mechanoreceptors, thermoreceptors, and nociceptors, some authors have proposed the existence of independent and parallel ascending streams that originate from different populations of rapidly or slowly adapting mechanoreceptors and terminate in the primary somatosensory cortex.29

Support for the view that these highly specialized parallel streams of tactile information underscore the existence of a strict labeled-line coding scheme comes from a variety of experimental observations. Implicit to this model is the assumption that local circuits within the intermediary relays of the somatosensory pathways, (the spinal cord, and several nuclei in the brainstem and thalamus) contribute little to the processing of ascending neuronal signals generated in the cutaneous periphery. According to this view, the main function of all subcortical relays of the somatosensory system is to faithfully transmit information sampled in the body's periphery to the neocortex, where all computations required for the emergence of a perceptual experience should take place. Thus, for the supporters of the FF model of touch, the observation of rapidly adapting (RA) or slowly adapting (SA) neuronal responses in the brainstem, thalamus, or even in the primary soma-tosensory cortex, is taken as direct evidence for the existence of segregated feedforward RA and SA pathways. Following this observation, the adaptation properties of central neurons should indicate whether they belong to either RA or SA pathways, which have their origins in the RA and SA fibers that innervate low-threshold mechanoreceptors in the animal's skin. Indeed, strict anatomical and physiological segregation schemes for RA and SA neurons have been proposed to exist in layer IV of area 3b of the primary somatosensory cortex of at least one primate species.118

The second important experimental observation commonly used to support the perceptual relevance of an FF model of touch is the finding that microstimulation of individual peripheral nerve fibers that innervate some classes of mechanoreceptors in humans can elicit distinct tactile perceptual experiences, which are often referred to as elementary sensations.122 These well-localized tactile sensations are experienced by subjects once the electrical stimulation of a given peripheral fiber reaches a critical threshold level. Electrical stimulation of single Meissner and Pacini units (i.e., fibers that innervate rapidly adapting mechanoreceptors) typically elicit a sensation of vibration, whereas similar stimulation of Merkel units (innervating a slowly adapting mechanoreceptor) can produce a sensation of sustained touch or pressure. Parametrical increase in the electrical stimulus intensity is often followed by the report of additional tactile sensations by the subject, since more and more tactile fibers are recruited by a stronger stimulus.56 Indeed, further increase in stimulus intensity can lead to paresthesias and even pain, likely because of recruitment of nociceptive fibers.

Additional experimental evidence clearly highlights the limits to which one can employ the findings described in the previous paragraph to support a strict FF view of the somatosensory system. First, no study to date has demonstrated the existence of a natural tactile stimulus capable of selectively activating either RA or SA mechanoreceptors. Instead, what is observed in reality is that both classes of mech-anoreceptors tend to respond to commonly used tactile stimuli, particularly when there is relative motion between the manipulandum and the skin surface. Second, the assumption that local circuits, at each relay station of the somatosensory, are incapable of altering incoming afferent signals is clearly not supported by the experimental evidence. To disprove this rather simplistic assumption, one needs only to point out that reduction in local inhibitory feedback, such as obtained by local infusion of GABA antagonists, leads to significant physiological changes in cortical and subcortical somatosensory neurons,66 which include changes in firing properties, such as response adaptation, and enlargement of neuronal receptive fields. In fact, local changes in the inhibitory tone or even in afferent-driven inhibition are believed to in part account for immediate receptive field reorganization that can be induced in the brainstem, thalamus, and cortex following a peripheral deafferentation.33 Thus, one can argue that the interplay of multiple local and extrinsic afferents that converge on brainstem, thalamic, and cortical neurons, as well as the peculiar intrinsic biophysical properties of these somatosensory neurons, are likely to influence the firing adaptation properties of these cells. We conclude that the presence of RA (phasic) and SA (tonic) tactile responses across the somatosensory pathway cannot be used as the sole criterion to infer that there are segregated feedforward labeled lines originating from each of the categories of mechanoreceptors. Phasic (RA) tactile responses are much more commonly encountered in central relays of the somatosen-sory system (ranging from 60 to 95% of the neurons) than in peripheral somatosensory fibers (62 to 71% of the single first order fibers).29 Moreover, RA neurons are also more often identified than SA neurons in the central nervous system.29 Since the disparity in the frequency of RA and SA neurons along the somatosensory system is usually higher than that observed in peripheral nerves, one can argue that local circuit interactions, such as the ones provided by local inhibitory feedback, may play a role in converting some of the original SA afferent signals into RA neuronal responses.

The nature of ascending somatosensory responses may also be altered by other modulatory pathways, such as the noradrenergic, serotoninergic, and cholinergic afferents that converge at each relay station of the somatosensory system. It is also likely that cortical feedback projections (see below) could contribute for the transformation of SA into RA responses in subcortical structures simply by modulating the inhibitory tonus provided by local interneurons.

As mentioned above, the elegant results from the microneurography studies in humans have often been used as the most decisive evidence in favor of the labeledline coding scheme of touch. However, though they may look compelling at first glance, there are several caveats that diminish their relevance as evidence in favor of a pure feedforward theory of touch. First of all, it is not surprising that stimulation of somatosensory fibers leads to some type of tactile sensation. The key question is whether the elementary sensations reported in these studies bear any resemblance to the actual percepts experienced by subjects engaged in active tactile discrimination tasks. Thus, the first concern one may have in interpreting the evidence generated in these experiments is the validity of using such an artificial stimulus (electrical microstimulation of single fibers) to categorize the perceptual capabilities of human subjects. It is safe to say that the human somatosensory system did not evolve to experience a single fiber stimulus and, as a consequence, the perceptual experiences elicited by such an uncommon stimulus should be far from the norm. The experimental evidence actually supports this prediction.121 At a critical level of microstimulation that presumptively activates only one tactile fiber, subjects report feeling a well localized tactile sensation, which resembles the original receptive field of the stimulated fiber. However, the same subjects often report that during this singlefiber stimulation they experience an odd, almost exotic mechanical sensation, which seems very unusual to them.121 Indeed, these subjects cannot relate this sensation to any real mechanical stimulus that they normally experience in real life.121 Subjective accounts like these support the observation that no natural tactile stimulus known to somatosensory physiologists is capable of selectively activating just a subpopulation of cutaneous mechanoreceptors, let alone a single somatosensory fiber. This is an important issue, particularly if one realizes that only a fraction of the single fibers that are stimulated electrically can elicit very distinct perceptual experiences.121 Indeed, stimulation of a considerable number of Ruffini units and even a subpopulation of Meissner, Merkel, and Pacini units produces no elementary tactile sensations whatsoever. Although the relevance of these negative results have been downplayed over the years, one only needs to examine them under a different framework to find support for the view that normal tactile perception emerges not by the transmission along a labeled line but rather through the integration across multiple ascending, as well as descending pathways. The fact that the elementary sensations produced by microstimulation of individual mechanoreceptor fibers are rather simple and unusual clearly distinguishes them from the type of complex, but familiar, types of tactile perceptual experiences that subjects experience when faced with real life tactile stimuli.

Another important point that is often neglected is the fact that the elementary sensations reported in the microstimulation studies result from a passive delivery of the tactile stimulus. Although this is not an issue for most proponents of the FF model of touch, those who defend a more interactive model of tactile perception would emphasize the integral role active movement contributes to the emergence of tactile percepts. Ethological evidence supports the notion that natural tactile perception emerges as the result of active exploration, which normally requires the engagement of the cutaneous periphery in manipulative behaviors that allow animals to actively scan the attributes of tangible objects. Although animals can perceive stimuli passively delivered to tactile organs, it is known that tactile discrimination occurs through the use of stereotyped behaviors, such as hand movements in primates and whisking in rodents. Since movement is known to modulate the activity of soma-tosensory neurons at all levels of the neuroaxis (see below), one would predict that engagement in an active tactile discrimination task would likely change the nature of the elementary sensations produced by microstimulation of single afferent fibers. In summary, even though one cannot deny the fact that electrical stimulation of individual somatosensory fibers elicits a conscious experience that can be described as an elementary tactile sensation, the role of these sensations in generating normal tactile percepts is far from clear. Certainly, viewed under the prism that we favor, the occurrence of the production of elementary tactile sensations by microstimulation of single peripheral tactile fibers sheds little light on whether labeled lines really play the dominant role in normal tactile perception that the proponents of the FF model of touch have postulated.

On their way to the neocortex, somatosensory pathways make synapses in a series of subcortical nuclei in the spinal cord, brainstem, and thalamus.62 In each of these intermediary relays, as well as in the somatosensory cortex, one can readily identify a topographic representation of the animal's body surface. An important feature of these maps is that they are somewhat distorted, since more neuronal tissue is used to represent body regions with high densities of low-threshold mechanore-ceptors (such as the hand in humans and non-human primates, and the whiskers in rodents). In each species, the distortion of these somatotopic representations also reflects the fact that body regions with high densities of mechanoreceptors invariably constitute the most important tactile organ used by animals for active tactile discrimination (e.g., hands and peri-oral regions in primates, whiskers in rodents, etc.).

Although one cannot deny the conspicuous presence of topographic maps of the body surface in every mammalian species, the precise role of topography in tactile perception is still open for debate.102 Indeed, the formulation of the main tenants of the FF theory of touch precedes the discovery, two decades later, that these maps are highly dynamic structures that can undergo considerable plastic reorganization throughout life.59 The plastic potential of both cortical and subcortical representations in the adult somatosensory system, which was not predicted by the FF model of touch, can no longer be ignored by any model of touch. This omission is particularly egregious if one takes into account new experimental evidence suggesting that plastic reorganization of cortical and thalamic maps, following limb amputation in humans, may account for the occurrence of a vivid tactile illusion known as phantom limb sensation.103 In its most perverse and paradoxical form, this phantom limb sensation can be accompanied by excruciating chronic pain in a part of the body that no longer exists.63 While information about location of a tactile stimulus could be coded and read out by stacks of topographic maps, it is important to emphasize that the lack of a precise somatotopic representation does not preclude information about stimulus location being extracted from populations of neurons located in a cortical area. For example, recent multi-electrode recordings in primates have revealed that information about stimulus location can be readily extracted, on a single-trial basis, from ensembles of neurons located in the secondary somatosen-sory cortex (SII) and area 2 of the parietal cortex, two regions in which one observes much less well-defined topographic maps than in the primary somatosensory (SI) cortex.92 Interestingly, due to a degree of overlap in the timing of SI, SII, and area 2 tactile responses, stimulus location could be derived almost simultaneously in all three cortical areas. In other words, the location of a tactile stimulus can be resolved by populations of somatosensory neurons, which define highly distributed represen-tations.92

We believe that in the process of defining a new theory of touch, the potential physiological role of somatotopic maps in general, as well as other modular neuronal structures, such as the barrels, barreloids, and barrelets that are observed throughout the trigeminal system of rodents, will have to be revisited. We tend to favor the notion that topographic maps primarily reflect the result of the self-organizing process that is responsible for wiring up the somatosensory system during early stages of development. Thus, although somatotopic representations may impose important constraints on the type of dynamic interactions and encoding schemes that can be implemented at certain levels of the somatosensory system, they do not necessarily preclude the existence of other representation schemes, even at the level of the thalamus and primary SI cortex. Recent experimental evidence suggests that the temporal domain of tactile neuronal responses, synchronous firing, and correlated neuronal activity could also play a role in tactile information processing38-85 that is independent of the topographic relationships.

Peripheral Neuropathy Natural Treatment Options

Peripheral Neuropathy Natural Treatment Options

This guide will help millions of people understand this condition so that they can take control of their lives and make informed decisions. The ebook covers information on a vast number of different types of neuropathy. In addition, it will be a useful resource for their families, caregivers, and health care providers.

Get My Free Ebook


Post a comment