How does the somatic sensory system contribute to
object manipulation and surface feature analysis?

Greg Detre

Friday, 12 October, 2001

Dr Iles, Neuro II, week 1

 

The somatic sensory system covers the four broad modalities of:

discriminative touch � the size, shape, force of contact, texture and movement of objects across the skin

proprioception � sensing the static and moving positions of the limbs and body

nociception � signalling tissue damage, perceived as pain (with different types of pain signalling different sorts of damage and response)

temperature sense � warmth and cold

Of these, it is our senses of touch (conveyed through four main types of mechanoreceptor located in various positions and depths in the skin), and to a lesser extent, proprioception, that provide a tactile perception of an object and tell us enough about it to manipulate it.

 

As with all neurophysiological investigations, there are various ways to study the somatic sensory system, which vary according to the technology, where in the body measurements are being taken from, and ethical issues. Single-unit (and more recently, multi-unit) recording in animals has been vital for progress, with rodent and primate systems being most studied. Study of human neurophysiology usually relies on non-invasive techniques like fMRI, although neuroimaging spatial resolution is still too low to really look at the columnar organisation of the cortex, for example. Psychophysical investigations record from single primary afferent fibres in awake human subjects using transdermal microneurography, first used by Karl-Erik Hagbarth and Ake Vallbo. Using this method, Vallbo showed that a subject�s reported sensory threshold can be as low as the receptive threshold of just one afferent fibre. Lesioning information can be useful, both in animals and in humans, especially for establishing localisation of function within the cortex.

The dorsal column-medial lemniscal system is the main pathway mediating tactile sensation and limb proprioception. Afferent mechanoreceptors in the skin (discussed below) are innervated by peripheral axons of nerve cells in the dorsal root ganglia. Their central branches ascend in the dorsal columns and synapse with second-order neurons in the dorsal column nuclei, which then cross the midline through the gracile and cuneate nuclei in the medulla and ascend through the brain stem on the contralateral side as the medial lemniscus. In the thalamus they synapse on third-order cells in the ventral posterior medial and ventral posterior lateral nuclei. This is the last subcortical waystation before projecting to the somatosensory cortical regions in the post-central gyrus of the parietal lobe.

There are four cutaneous mechanoreceptors that innervate the glabrous skin (with the highest density in the fingertips). Slowly-adapting mechanoreceptors respond continuously to a persistent stimulus, whereas rapidly-adapting mechanoreceptors signal a change - they respond initially, but will have ceased to fire after a few hundred milliseconds even if the stimulus continues, although they sometimes fire when the stimulus terminates too.

The superficial skin has slowly adapting Merkel�s cells and rapidly adapting Meissner�s corpuscles. The deeper tissue contains rapidly adapting pacinian corpuscles and slowly adapting Ruffini�s corpuscles. The actual mechanosensitive transducers reside in the unmyelinated endings of afferent fibres, with the receptors� selectivity seeming to depend as much on the receptor structure surrounding the transducer endings as the endings themselves.

Merkel�s disks have a simple structure. They are positioned in the basal layer of the epidermis, and are particularly densely located in the fingertips (about 100 per cm2). They have a set of properties that make them ideal for providing detailed information about the surface texture and curvature of a tactile stimulus: they respond to sustained indentation with a sustained, slowly-adapting discharge that is linearly related to indentation depth; are highly sensitive to points, edges and curvature; are able to resolve spatial detail of 0.5mm (although their receptive field diameters are 2-3mm); are hardly affected by the contact force of the stimulus; and are at least ten times more sensitive to dynamic than static stimuli (at up to 80mm/s).

Meissner corpuscles are relatively large cell assemblies in the dermal ridges that lie just beneath the epidermis. Their pillow-like arrangement appears to act as a filter protecting the endings of their rapidly-adapting afferent fibres from static skin deformation. They are even more sensitive to dynamic stimulation (objects or surfaces moving over the skin) than the slowly adapting Merkel�s cells. Because they respond to stimuli over their entire receptive fields, they resolve spatial detail poorly. However, this makes them ideal in grip control for signalling the frequent microscopic slips between object and skin which evoke reflexive increases in grip force.

The Ruffini corpuscle is located in the connective tissue of the dermis is a relatively large spindle shaped structure tied into the local collagen matrix (like a Golgi tendon organ in muscle). They innervate the skin less densely than Merkel�s or Meissner�s corpuscles. Because they have not been observed in neurophysiological studies of mechanoreceptors in the monkey hand, they have been studied less extensively than the other afferent types. Their receptive fields are about five times larger, six times less sensitive to skin indentation but 2-4 times more sensitive to skin stretch than Merkel�s corpuscles.

Johnson identifies two important roles for Ruffini corpuscles:

1.       Perception of the direction of object motion/force when producing skin stretch (although Gardner and Sklar have shown with vibrating pins that activate only Meissner�s corpuscles and Pacinian afferents that motion and motion direction are still discriminated effectively without Ruffini afferent input).

2.       Assisting (along with muscle spindles and possibly joint afferents) in the perception of hand shape and finger position through the pattern of skin stretch produced by each hand and finger conformation - simply stretching this skin, activating Ruffini afferents strongly (and Merkel�s corpuscles afferents more weakly) produces the illusion of finger flexion, as does tendon vibration

The deep location of the Ruffini corpuscles seems to shield it from the effects of indentation. Johnson speculates that this greater sensitivity to stretch than indentation means greater sensitivity to horizontal tensile strain (shear), which helps deduce the object�s direction of motion and hand conformation.

Pacinian corpuscles are remarkable sensitive, sometimes responding to 10nm of skin motion (about two orders of magnitude more sensitive than Meissner�s corpuscles. Consequently, they have multiple, fluid-filled, cushioning layers that filter out the low-frequency stimuli that would otherwise overwhelm them. They have extremely low spatial resolution, perhaps even having the entire hand as a receptive field.

Because they are rapidly adapting, they provide a high-fidelity signal of transient and vibratory stimuli. Indeed, they are sensitive enough to allow us to perceive distant events through transmitted vibrations when we grasp an object in the hand (Hunt in Johnson, Yoshioka and Vega-Bermudez, 2000). Perhaps even more importantly, they are responsible for perceiving events at the working surface of a tool/probe that we are skilled with, as though our fingers were present at the end.

Merkel�s corpuscles can be found in hairy skin too, along with Pacinian and Meissner�s corpuscles, although they are usually slightly differently formed. However, the principle mechanoreceptor in hairy skin is the hair follicle, of which there can sometimes be three types: guard, tylotrich and down. These are obviously less relevant to object manipulation, since they only really signal an uninformative flutter.

 

There are three major divisions to the somatic sensory cortex: the primary (S-I) and secondary (S-II) somatosensory cortices and the posterior parietal cortex. S-I is cytoarchitecturally subdivided into Brodmann�s areas 1, 2, 3a and 3b. Most of the thalamic fibres terminate in 3a and 3b, although there are also sparser projections directly to 1 and 2, to S-II and to the posterior parietal cortex, or somatosensory association cortex (Brodmann�s areas 5 and 7).The S-I cells in 3a and 3b project mainly to 1 and 2, but also to the posterior parietal cortex and to the symmetrical areas of the S-I in the opposite hemisphere (ipsilateral to the original primary afferents). S-II is mainly innervated by neurons from each of the four areas of S-I. Indeed, if the neural connections in S-I representing, say, the hand area are completely removed, this completely prevents stimuli applied to the skin of the hand from activating neurons in S-II (Pons et al. 1987). S-I cells in layer 4 are primarily interneurons that connect with other cortical layers, while those in layers 5 and 6 project back to the thalamus and to the striatum).

Notably, the structure of the somatosensory cortex is highly organised, somatotopically (i.e. a topographic representation of the body is maintained) and in columns according to modality. There are two main types of cells in the somatosensory cortex: the pyramidal (output) and non-pyramidal (within layers). The sensory homunculus (Penfield and Rasmussen, 1950) illustrates in grotesque graphic form which areas of the body are most highly represented in the cortex, and have correspondingly high densities of receptors (e.g. tongue, lips, fingertips, hand).

The functionality of S-I follows cytoarchitectural lines:

3a stimulation from muscle afferents

3b stimulation from Merkel's disc and Ruffini organs (slowly-adapting)

1 stimulation from Meissner's and pacinian corpuscles (rapidly-adapting)

2 stimulation from joint afferents

The posterior parietal cortex relates sensory and motor processing, and integrates the different somatic sensory modalities, from S-II and the thalamus. It appears to function in perception of body parts in relation to one another, body parts in relation to the external environment, relation of objects in the environment, spatial memory, and attention.

Each of the four mechanoreceptors respond to cutaneous motion and deformation in a different way, together giving rise to our complete sense of tactile perception. Johnson (2001) further suggests that �the distinctively different functions identified for the four cutaneous mechanoreceptive afferent systems suggest the existence of distinct and separate central systems for processing the information provided by each of the primary afferent groups�, arguing that many of the computational problems for form and texture can be separated from motion and motion direction. He gives the example of neurons in 3b, which are highly selective for spatial form and have mechanisms for preserving spatial information at high scanning velocities, but are no more sensitive to motion or motion direction than are primary afferents (DiCarlo & Johnson, 2000).

The somatic sensory system provides a wealth of tactile data which is processed through the thalamus, S-I, S-II and the posterior parietal cortex. At each level, wider information is integrated, from the other side of the body, larger receptive fields, the different modalities, and eventually merged with visual data necessary for motor output.