Early Touch: From Neural Coding to Haptic Space Geometry

CNS*2013 Paris Workshop, July 17

Ehud Ahissar

Motor-sensory contingencies and morphological coding of objects 

In the rat vibrissal system, mechanoreceptors situated in the whisker follicle transduce signals that are initially processed mechanically by the external shaft of the whisker. I will present data showing that during active touch the mechanical responses of the whisker shaft contain morphological and dynamic representations of object location. The radial and azimuthal coordinates of the position of the contacted object are reliably represented in phase planes spanned by pairs of morphological variables of the whisker. Measurements in head-fixed and freely-moving behaving rats showed that rats apply forces that allow meaningful morphological coding, that coding is reliable as fast as 15 ms after contact, and that localization accuracy in behaving rats is correlated with code availability. This study does not show how these codes are read out, but shows that mechanical transformation from the whisker to the follicle is rigid and suggests possible readout mechanisms. I will discuss the relationships between morphological coding and motor-sensory contingencies.

  • Bagdasarian K, Szwed M, Knutsen PM, Deutsch D, Derdikman D, Pietr M, Simony E and Ahissar E (2013). Pre-neuronal morphological processing of object location by individual whiskers. Nature Neuroscience, 16(5):622-31, doi: 10.1038/nn.3378.
  • Saig A, Gordon G, Assa E, Arieli A and Ahissar E (2012). Motor-sensory confluence in tactile perception. Journal of Neuroscience, 32(40):14022-32, doi: 10.1523/JNEUROSCI.2432-12.2012.

Angelo Arleo

Early processing and coding of fine touch percepts 

Fine touch sensing relies on peripheral-to-central neurotransmission of somesthetic percepts. Here, we first focus on the encoding/decoding of primary tactile signals (fingertip mechanoreceptor responses) recorded via human microneurography experiments. We show that first-spike latencies of mechanoreceptor responses carry enough information for fine touch discrimination within a few milliseconds of the first afferent spike. Also, the relative timing of entire spike trains encode isometric representations of the stimulus space, a likely basis for generalisation in haptic perception. Second, we consider the propagation of primary signals up to cuneate neurons in the brainstem. We model cuneate responses based on in vivo intracellular recordings from non-anaesthetised but decerebrated cats. We modulate the efficacy of mechanoreceptor-cuneate synapses by spike-timing-dependent plasticity. Our simulations predict that optimal information transfer at the cuneate level occurs when only 4-8 synapses are non-silent, which is verified by in vivo recordings in cats. Third, we set forth a neuroengineering framework to design a closed-loop robotic system for fine touch sensing. The robotic implementation takes inspiration from known multistage neural coding mechanisms along the ascending somatosensory pathway. It uses a probabilistic algorithm for online tactile recognition downstream from the cuneate network. Finally, it implements active sensing through adaptive motor control. We probe the robotic system in a Braille reading task. Results on peripheral encoding of primary contact features are consistent with human SA-I mechanoreceptor responses. During Braille reading, 2nd-order processing by simulated cuneate neurons helps resolving perceptual ambiguities by increasing the separability of primary afferent signals. As a consequence, online discrimination of Braille characters is fast and fairly accurate.

  • Bengtsson F, Brasselet R, Johansson RS, Arleo A and Jorntell H (2013). Integration of sensory quanta in cuneate nucleus neurons in vivo. PLoS ONE, 8(2):e56630.
  • Brasselet R, Johansson RS and Arleo A (2011). Quantifying neurotransmission reliability through metrics based information analysis. Neural Comput 23(4):852-81.

Sliman Bensmaia

Spatial and temporal codes mediate the tactile perception of natural textures

When we run our fingers over the surface of an object, we acquire information about its microgeometry and material properties. Texture information is widely believed to be conveyed in spatial patterns of activation evoked across one population of cutaneous mechanoreceptive afferents. Here, we show that a spatial mechanism can only account for the processing of coarse textures. Information about most natural textures, however, is conveyed through high-precision temporal spiking patterns in afferent responses, driven by skin vibrations elicited during scanning. Furthermore, these texture-specific spiking patterns dilate or contract systematically in time with changes in scanning speed, suggesting a mechanism to achieve perceptual constancy across speeds. Our results imply a novel mechanism for encoding surface texture, one that implicates vibration-sensitive mechanoreceptive afferents and draws strong analogies with auditory and vibrissal processing.

  • Harvey MA, Saal HP, Dammann JF and Bensmaia SJ (2013). Multiplexing stimulus information through rate and temporal codes in primate somatosensory cortex. PLoS Biology, 11, e1001558.
  • Mackevicius EL, Best MD, Saal HP and Bensmaia SJ (2012). Millisecond precision spike timing shapes tactile perception. Journal of Neuroscience, 32, 15309-17.

Vincent Hayward

How the mechanical properties of the fingertip may impact the tactile sensory function

It can be surmised that the early sensory neural processes do not operate in the abstract, but in a manner that is grounded in the mechanics of the finger and of its interaction with objects. The fingertip and the glabrous skin have surprising properties that determine the occurrence of strongly nonlinear and nonlocal transformations between the attributes of objects and the mechanical state of the tissues in which the mechanoreceptors are embedded. Such attributes include object geometry, their resistance to deformation, or their microstructure. The study of the tribology of the skin and of its consequences on its interaction with objects reveals a dominant role of water in these transformations, with likely consequences in the sensory and manipulative functions of the hand. It is not clear whether these complexities represent challenges to be contented with or whether they provide the brain with unsuspected manipulative and sensory opportunities.

  • Wiertlewski M, Hayward V (2012). Mechanical behavior of the fingertip in the range of frequencies and displacements relevant to touch. Journal of Biomechanics, 45(11):1869-1874.
  • Andre T, Levesque V, Hayward V, Lefevre P and Thonnard J-L (2011). Effect of skin hydration on the dynamics of fingertip gripping contact. Journal of the Royal Society Interface, 8(64):1574-1583.

Jan Koenderink

Assemblage and Icon in Perception and Art

In his influential The Problem of Form (1893), the German sculptor Adolf Hildebrand distinguishes categorically between perception obtained from multiple fixations or vantage points (G. Bewegungsvorstellungen - I call these "assemblages"), and from purely "iconic" imagery (G. Fernbilder). Only the latter he considers properly "artistic". Hildebrand finds the reason for his ontological distinction in the microgenesis of perceptual awareness.
What to make of this?
I identify two distinct kinds of assemblage, those that involve distinct perpectives, and those that involve mere fixation differences. There are marked empirical differences between the three kinds of perceptual awareness. Assemblages involving perspectives are "motor images", iconic perceptions "glued together" through movements. This is akin to haptic awareness (indeed, so described by Hildebrand). The awareness equally involves the sense and action worlds. Assemblages involving fixations lead to marked non-veridical spatial impressions, that are remarkably similar in the case of haptic and optical observations. The nature of the "glue" is peculiar. It has a "self repairing" character that remains ill understood. Finally, icons appear to be purely visual entities. They have different properties (often denoted "pictorial") from generic vision, which is fully dominated by assemblages. Pictorial space is a 2+1D fiber bundle, whereas the assemblages approximate Euclidean 3-space. Binocular stereo represents a curious, in-between case, involving a virtual motion.

Masashi Nakatani

Tactile geometric perception explained by contact mechanics

We can readily judge whether the contact surface is flat, raised, or indented. It seems to be easy task, but what is the neural mechanism behind? This is a question worth pursuing because it can give some insights how we interpret the geometric information from tactile afferent input. To tackle this problem, we've developed the finite element model of fingerpad that can simulate the deformation produced by geometric surface. This mechanics model can estimate the 2D distribution of stress concentration produced by geometric stimuli (flat, raised or indented shape) in the location where mechanoreceptors are frequently observed. By interpreting the simulated result, we proposed a mechanism how people can judge the geometric shape contacting to the fingerpad. In the middle of the talk, we also introduce the phenomenon "Fishbone Tactile Illusion", which produces the perception of illusionary indented geometry even though the surface is physically flat. This phenomenon is intriguing because it suggests the interaction between geometric stimulus and corse texture encoding in afferent nerves. We discuss the possible explanation how this phenomenon occurs in terms of the functional characteristic of mechanoreceptors and their innervating afferent nerves.

  • Nakatani M, Howe RD and Tachi S (2011). Surface texture can bias tactile form perception. Experimental Brain Research, 208(1):151-156.
  • Nakatani M, Sato A, Tachi S and Hayward V (2008). Tactile illusion caused by tangential skin strain and analysis in terms of skin deformation. Haptics: Perception, Devices and Scenarios, Lecture Notes in Computer Science, 5024:229-237.

Tony Prescott

Does tactile attention make use of salience maps?

Sensory input is strongly dependent on where attention is directed.  However, despite its critical role in determining experience, research on attention has largely focused on the visual modality due to the comparative ease with which eye movements can be measured; the role of attention in touch is comparatively neglected.   This talk will propose the rodent whisker system as a useful model system in which to study attention in the modality of touch. Several reliable measures of whisker movements have now been developed, and related to environmental and behavioral variables, making it possible to use whisker movement and positioning as clues to the animal’s attentional focus. Here, we propose a computational model of whisker movement considered as being driven by attentional mechanisms, and show, using computer-simulated behavioural experiments, that this model is consistent with a broad range of experimental observations. A core hypothesis is that the rat explicitly decodes the location in space of whisker contacts and that this representation is used to regulate whisker drive signals. This proposition stands in contrast to earlier proposals that the modulation of whisker movement during exploration is mediated primarily by reflex loops. The current model is abstract in form, however, we argue that the superior colliculus is a candidate neural substrate for the implementation of a unified head-centred tactile salience map, in analogy to current models of (primate) visual attention. The proposed model has the potential to offer both a more complete understanding of sensory processing in touch as well as providing an experimental tool for the study of attention. (Talk based on ongoing research in collaboration with Ben Mitchinson).

Elie Wandersman

Fine texture discrimination : a biomimetic approach

We present a biomimetic approach of tactile perception, focused on the fine texture discrimination capabilities. We study in this purpose the friction between a roughened glass slide and an elastomer spherical cap whose surface is patterned with parallel grooves of spatial period L ~100 µm, mimicking human fingerprints. For a prescribed normal load P (20 mN – 2 N) and velocity v (0,01 – 1 mm/s), the  tangential friction force Q exhibits minute oscillations around its time-averaged value. The normalized spectrum of Q reveals the existence of a dominant mode centered at frequency v/L, whose amplitude (i)  strongly depends on the probed roughness characteristics and (ii) decreases analytically with P. A simple model is developed, based on the normal load dependence of the dynamical friction coefficient, which captures the mode selection at v/L and quantitatively accounts for the dependence of the oscillation amplitude with both the confining force P and the roughness characteristics of the glass slide. In this description, the periodicity of the elastomer surface topography induces a spectral amplification of the contact stress field, a mechanism that should be relevant to any frictional configuration involving regularly patterned surfaces and could be applied to robotic haptic devices.

  • Wandersmann E, Candelier R, Debrégeas G and Prevost A (2011). Texture-Induced Modulations of Friction Force: The Fingerprint Effect. Physical Review Letters, 107(16):164301-164305. 

Junji Watanabe

Time and motion in touch - psychophysical approach

In tactile processing, peripheral mechanoreceptors are distributed over the body surface and multiple types of mechanoreceptor channels transmit signals to the brain. How the brain integrate signals from different bodily sites and different channels is a foundamental issue of tactile perception. Here we investigate  the computational mechanism of time and motion perception in touch by using selective adaptation paradigm.

  • Kuroki S, Watanabe J and Nishida S (2013). Contribution of within- and cross-channel information to vibrotactile frequency discrimination. Brain Research, 1529:46-55.