INVESTIGATION OF AN ADAPTATION-INDUCED TACTILE SPATIAL ILLUSION INVESTIGATION OF AN ADAPTATION-INDUCED TACTILE SPATIAL ILLUSION: PSYCHOPHYSICS AND BAYESIAN MODELING By LUXI LI, Hon. B.Sc. A Thesis Submitted to the School of Graduate Studies in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy McMaster University © Copyright by Luxi Li July 2017 Ph.D. Thesis – Li, L. McMaster University - Psychology ii McMaster University DOCTOR OF PHILOSOPHY (2017) Hamilton, Ontario (Psychology) TITLE: Investigation of an adaptation-induced tactile spatial illusion: psychophysics and Bayesian modeling AUTHOR: Luxi Li, Hon. B.Sc. (McMaster University) SUPERVISOR: Dr. Daniel Goldreich NUMBER OF PAGES: xiv, 128 Ph.D. Thesis – Li, L. McMaster University - Psychology iii Lay Abstract Sensory adaptation can shape how we perceive the world. In this thesis, we showed that the perception of space in touch is pliable and subject to the influence of adaptation. Psychophysical testing in human participants showed that vibratory adaptation induced an illusion that expanded the perceived distance between stimuli on the skin. This illusion provides clues into how information about space in touch is normally processed and interpreted by the brain. In addition, we developed a computational model that used a powerful statistical framework – Bayesian inference – to investigate touch on a theoretical basis. To the best of our knowledge, the present thesis provides the first combined psychophysical and computational study on the effects of adaptation on tactile spatial perception. Our findings suggest that touch shares some common information processing principles with vision and hearing, and adaptation plays a functionally similar role in mediating this process across the senses. Ph.D. Thesis – Li, L. McMaster University - Psychology iv Abstract Sensory adaptation is an important aspect of perception. A seemingly non-beneficial consequence of adaptation is that it produces perceptual illusions. For instance, following focal adaptation, the perceived separation between stimuli straddling the adapted attribute or region is often exaggerated. This type of illusion, known as perceptual repulsion, is both a consequence of and a clue to the brain’s coding strategies and how they are influenced by recent sensory events. Adaptation-induced perceptual repulsion has been well documented in vision (e.g. the tilt aftereffect) and to a lesser extent in audition, but rarely studied in touch. The present thesis investigated the effects of adaptation on tactile spatial perception using a combination of human psychophysics and computational modeling. In a two-interval forced choice task, participants compared the perceived separation between two point-stimuli applied on the forearms successively. The point of subjective equality was extracted as a measure of perceived two-point distance. We showed that tactile spatial perception is subject to an adaptation-induced repulsion illusion: vibrotactile adaptation focally reduced tactile sensitivity and significantly increased the perceived distance between points straddling the adapted skin site (Chapter 2). This repulsion illusion, however, was not observed when the intervening skin was desensitized with topical anesthesia instead of vibrotactile adaptation, suggesting that peripheral desensitization alone is insufficient to induce the illusion (Chapter 3). With Bayesian perceptual modeling, we showed that the illusion was consistent with the hypothesis that the brain decodes tactile spatial input without awareness of the adaptation state in the nervous system (Chapter 4). Together, the empirical and theoretical work furthers the understanding of dynamic tactile spatial coding as the somatosensory system adapts to the sensory environment. Its main findings are consistent with the adaptation- induced repulsion illusions reported in vision and audition, suggesting that perception in different sensory modalities shares common processing features and computational principles. Ph.D. Thesis – Li, L. McMaster University - Psychology v Preface This thesis is composed of five chapters. Chapter 1 overviews the background literature in tactile perception, sensory adaptation, and perceptual illusions, and introduces the reader to Bayesian ideal observer analysis. Chapters 2 and 3 are empirical studies using psychophysical experimentation in human participants. Chapter 2 is published in the open-access journal Frontiers in Human Neuroscience 1, and is permitted for inclusion in the thesis under the terms of the Creative Commons Attribution License2. Chapter 4 is a computational modeling study exploring the empirical results described in Chapters 2 and 3. Chapter 5 discusses the findings and implications of these studies. The research detailed in this thesis was supported by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant awarded to Dr. Daniel Goldreich. It was also supported through an annual graduate stipend from 2011 to 2016 by McMaster University. 1 Li, L., Chan, A., Iqbal, S. M., & Goldreich, D. (2007). An adaptation-induced repulsion illusion in tactile spatial perception. Front. Hum. Neurosci. 11: 331. doi: http://10.3389/fnhum.2017.00331 2 https://creativecommons.org/licenses/by/4.0/ Ph.D. Thesis – Li, L. McMaster University - Psychology vi Declaration of Academic Achievement Chapter 2 I was involved in all aspects of this empirical research: experimental design, programming, data collection, statistical analysis, and writing. My graduate supervisor Dr. Goldreich made major contributions to the experimental design and programming. An undergraduate thesis student, Shah Mehran Iqbal, assisted in the data collection and statistical analysis for the pilot phase of the research (not included in this thesis). Another undergraduate thesis student, Arielle Chan, assisted in the data collection for the pilot phase and for Experiments 2 and 3. Several undergraduate laboratory study students also assisted in the data collection: Soumya Saini, Kyle Gauder, Vy Ngo, Faiza Shafaqat, Hiral Patel, Kaitlyn Gonsalves, and Cecelia Dai. Chapter 3 I was involved in all aspects of this empirical research: experimental design, programming, data collection, statistical analysis, and writing. Dr. Daniel Goldreich contributed to the experimental design and provided guidance at every stage of the research. An undergraduate thesis student, Jessica Webber, assisted in the data collection and statistical analysis. Several undergraduate laboratory study students also assisted in the data collection: Tyler Czaniecki, Cecelia Dai, Adam Alic, Jenin El-Sayes, Shirley Ong, and Claudia Turco. Chapter 4 I was involved in all aspects of this modeling research: conceptualizing, programming, running simulations, collecting data, performing statistical analysis, and writing. Dr. Daniel Goldreich provided guidance at every stage, and made major contributions to the conceptualizing and programming. An undergraduate thesis student, Michael Wan, assisted in running simulations and collecting data. Ph.D. Thesis – Li, L. McMaster University - Psychology vii Acknowledgements First and foremost, I would like to thank my graduate supervisor, Dr. Daniel Goldreich, for his guidance, support, and encouragement all these years. Dan is a wonderful mentor. His scientific enthusiasm is contagious. His devotion to research and education is inspiring. He is the type of researchers who find fulfillment in the purest pursuit of scientific discovery. He is also one of the nicest and most patient people I have met. I would like to extend my gratitude to my committee members, Drs. Deda Gillespie, Patrick Bennett, and David I. Shore, for their time, invaluable advice, and kindness that guide me through this rite of passage. I would also like to thank Dr. Jeffrey Min-in Yau, for his thoughtful feedback on this dissertation and his advice on finding the personal path in academia and science. Some of my happiest moments in grad school were thanks to the company of my former lab mates, Drs. Andy Bhattacharjee, Jonathan Tong, Ryan Peters, and particularly, Mike Wong. I miss our laughter and little adventures together. I would also like to thank my current lab mates, Arnav Bharadwaj, Kyle Gauder, Akash Deep, and Nina Prodribaba, for continuing the merry lab spirit and their helpful comments on my manuscript. My experimental work would not have been possible without the assistance of many undergraduate students over the years, to whom I am grateful: Arielle Chan, Jessica Webber, Tyler Czaniecki, Shah Mehran Iqbal, Michael Wan, Soumya Saini, Cecelia Dai, Vy Ngo, Faiza Shafaqat, Hiral Patel, Adam Alic, Jenin El-Sayes, Kaitlyn Gonsalves, Shirley Ong, and Claudia Turco. During my PhD and life journey in Canada, many friends have lent me an ear and perked me up when I felt down. Their emotional support helped me push through. In addition to my lab mates, I would like to give specific thanks to Savitri Jetoo and Alan Santinele Martino. There are too many other names to mention explicitly, but know that I thank you in my heart. In the past a few years, the person who has been the closest to me, sharing both joy and tears with me, is my partner Trevor Copp. Trevor, thank you for having my back. You make me a better person, and you make these all possible. After being a foreigner in Canada for 11 years, you are where I call home. I’ll extend my gratitude to Trevor’s lovely family: You are my family in Canada. I would like to dedicate this dissertation to my parents, Wang Jianmin and Li Zhencai, as well as my stepfather, Liang Hanzhen, for their love and support. Mom, although there were difficult times in the family as I grew up, I know that you love me deeply in the best way you know, and you did the best you could; for that I am infinitely grateful. Dad, thank you for giving me the freedom to explore what I want. Uncle Liang, thank you for being an incredible person and for your tremendous support for Mom and me. Ph.D. Thesis – Li, L. McMaster University - Psychology viii Table of Contents CHAPTER 1 1 GENERAL INTRODUCTION 1.1 An overview of neural mechanisms of tactile perception 1 1.2 Adaptation and perceptual illusions 5 1.3 Modeling perception as probabilistic inference 8 1.4 Overview of studies 9 1.5 References 10 CHAPTER 2 15 ADAPTATION 2.1 Preface 15 2.2 Abstract 16 2.3 Introduction 16 2.4 Materials and Methods 17 2.5 Results 21 2.6 Discussion 24 2.7 Conclusion 28 2.8 References 28 CHAPTER 3 31 ANESTHESIA Ph.D. Thesis – Li, L. McMaster University - Psychology ix 3.1 Preface 31 3.2 Abstract 32 3.3 Introduction 32 3.4 Methods 33 3.5 Results 39 3.6 Discussion 42 3.7 Conclusion 47 3.8 References 47 3.9 Figures and Captions 51 CHAPTER 4 58 COMPUTATIONAL MODELING 4.1 Preface 58 4.2 Abstract 59 4.3 Introduction 59 4.4 Methods 62 4.4.1 Encoding models 62 4.4.2 The Bayesian decoders 66 4.5 Results 72 4.6 Discussion 76 4.7 Conclusion 93 4.8 References 93 4.9 Figures and Captions 100 Ph.D. Thesis – Li, L. McMaster University - Psychology x CHAPTER 5 111 GENERAL DISCUSSION 5.1 Summary of studies 111 5.2 Possible cellular and synaptic mechanisms underlying tactile adaptation 112 5.3 Functional benefits of adaptation 115 5.4 Implications and future directions 117 5.5 Conclusion 121 5.6 References 121 Ph.D. Thesis – Li, L. McMaster University - Psychology xi List of Figures CHAPTER 2 Figure 1 Experimental protocols and expected results 18 Figure 2 Force-controlled two-point stimulus apparatus 19 Figure 3 Experiment 1 results 22 Figure 4 Experiment 2 results 24 Figure 5 Experiment 3 results 25 Figure 6 A model for adaptation-induced tactile spatial aftereffects 27 CHAPTER 3 Figure 1 Experimental protocols (left) and expected results (right) 51 Figure 2 Monofilament detection results 53 Figure 3 Two-point distance comparison results 54 Figure 4 Normalized threshold elevation by anesthesia vs. adaptation as a function of distance from centre of anesthesia or adaptation 55 Figure 5 A simple schematic model for adaptation vs. anesthesia effects on tactile spatial perception 56 CHAPTER 4 Figure 1 Simulated RFs and adapting stimulus 100 Figure 2 Modeling two-point distance comparison 101 Ph.D. Thesis – Li, L. McMaster University - Psychology xii Figure 3 Modeling monofilament detection 102 Figure 4 Top 12 best-fitting model performances in the no-adaptation (NA) condition 103 Figure 5 Effects of adaptation on neural population response in the encoder 105 Figure 6 Effects of adaptation via an aware vs. unaware decoder 106 Figure 7 Effects of adaptation on two-point distance perception indicated by psychometric function and PSE 107 Figure 8 Figure 9 Figure 10 Effects of adaptation on single-point detection Model performance on either perceptual task using the best- fitting parameters from the other task Illustrations for Discussion 108 109 110 Ph.D. Thesis – Li, L. McMaster University - Psychology xiii List of Abbreviations and Symbols Abbreviations 2IFC Two-Interval (two-alternative) Forced-Choice A or Adapt Adaptation condition A0, A3, A7 Adaptation condition with 0, 3, or 7 seconds of top-ups in between trials Anes Anesthesia condition CNS Central Nervous System
 DCML Dorsal Column Medial Lemniscus FA Fast-Adapting ips Impulses per second (firing rate unit) LLR Log Likelihood-Ratio LML Log Maximum Likelihood NA No-Adaptation or No-Anesthesia (baseline) condition PDF Probability Distribution Function PSE Point of Subjective Equality RA Rapidly-Adapting (afferent) RF Receptive Field RMSE Root Mean Square Error SD Standard Deviation SE Standard Error S1 Primary somatosensory cortex S2 Secondary somatosensory cortex 
 SA Slowly-Adapting Ph.D. Thesis – Li, L. McMaster University - Psychology xiv SA1 Slowly-Adapting type 1 (afferent) SA2 Slowly-Adapting type 2 (afferent) TAE (Visual) Tilt After-Effect V1 Primary visual cortex
 VPM Ventral Posterior Medial nucleus (in the thalamus) Symbols A Amplitude of Gaussian function; expected spike count at the receptive field centre c Expected (average) stimulus-evoked spike count C Comparison distance in a two-point distance-comparison task d Distance between receptive field centre and a stimulus k Actual spike count on a single trial R Reference distance in a two-point distance-comparison task s Stimulus strength x x-position in a simulated skin patch y y-position in a simulated skin patch Δx Distance between x-positions of two point-stimuli; two-point distance α Adaptation state λ Total expected spike count σ Standard deviation Ψ Psychometric function µ Mean γ y-intercept δ Lapse rate Ph.D. Thesis – Li, L. McMaster University - Psychology 1 CHAPTER 1 GENERAL INTRODUCTION 1.1 An overview of neural mechanisms of tactile perception Touch is essential in everyday life. It allows us to directly interact with the environment. It is through touch that we determine the surface properties of materials, manipulate and control objects, utilize tools to expand our reach and abilities, obtain accurate awareness of bodily position and movement, feel pleasure and pain, and establish connection and rapport with fellow humans by simple contact, such as a handshake or a pat on the shoulder. We often take for granted our perception of the world through touch, but how does the somatosensory system accomplish this remarkable feat? The quick answer is we know very little. Despite its ubiquitous presence and immense importance, touch has been underinvestigated compared to vision and audition. Even though the history of research on touch can be dated back to Aristotle (Fulkerson, 2016), it was not until about 50 years ago that the fascinating neural mechanisms underlying somatosensation began to be unravelled (Morley, 1998). Touch is more than skin deep. The somatosensory system comprises subsystems that are responsible for coding very different physical stimuli, such as pressure, vibration, temperature, pain, body position, and movement. This thesis is in the context of passive tactile perception, which mainly deals with the perception of light touch from stimuli imposed on the skin surface without active exploration. The lack of active exploration limits the proprioceptive and motor information. At first glance, passive tactile perception of simple stimulation such as point-pressure may appear to be a very basic task, but it is mediated by a complex and sophisticated system of neurons and pathways. Tactile perception begins with transduction in mechanoreceptors in the periphery, where the skin deformation caused by mechanical stimuli is converted into neural signals in the form of action potentials and transmitted by myelinated Aβ fibers (mechanoreceptive afferents), into the dorsal column medial lemniscus (DCML) pathway in the central nervous system (CNS) (Rustioni et al., 1979). Via this pathway and relayed by three orders of neurons, the signals travel through the dorsal column nuclei in the medulla, and the ventroposterior lateral nuclei in the thalamus, into the primary and secondary somatosensory cortices (S1 and S2, respectively). In addition to these thalamocortical projections of neural signals, there are extensive intracortical projections: S1 projects to S2, and both S1 and S2 project to downstream cortical areas (for reviews, see Iwamura, 1998; McGlone & Reilly, 2010; Bensmaia & Yau, 2011; Serino & Haggard, 2010). Ultimately, neural signals are translated into conscious perception. Ph.D. Thesis – Li, L. McMaster University - Psychology 2 An important component of tactile information processing is its spatial dimension. The extraction of information about many properties of an object, such as its size, shape, and surface texture, relies on identifying where and how the skin is stimulated. Simple spatial perception, such as locating a stimulus or determining the distance between two stimuli on the skin, is at the foundation of performing many complex tactile tasks. Here, the main research interest of the present thesis is tactile spatial perception. Specifically, this thesis examines how recent tactile experience influences tactile spatial perception. Spatial information about a tactile stimulus is first conveyed by peripheral tactile receptors. Much of the knowledge about tactile receptors has come from research on the glabrous (i.e. hairless) skin of the fingertips. This is because the fingertips possess a plethora of tactile receptors and are important for a wide variety of tactile tasks. Neurophysiological studies have identified four major classes of primary afferents and their associated mechanoreceptors in the glabrous skin of the fingertips in humans, non- human primates, and other mammals. These primary afferents are categorized and named based on the adaptation properties of their responses and their receptive field (RF) sizes. They are: slowly-adapting type 1 (SA1) afferents, innervating Merkel cells; slowly- adapting type 2 (SA2) afferents, innervating Ruffini endings; rapidly-adapting (RA) afferents, innervating Meissner corpuscles; and PC afferents, innervating Pacinian corpuscles. RA and PC afferents are also called fast-adapting type 1 and type 2 (FA1 and FA2) afferents respectively in the literature (Johansson & Flanagan, 2009). The slowly- adapting afferents (SA1 and SA2) fire throughout a sustained indentation, whereas the fast-adapting afferents (RA and PC) fire only at the onset and offset of the indentation. The RFs of type 1 afferents (SA1 and RA) are small and well-defined, whereas the RFs of type 2 afferents (SA2 and PC) are large with borders that are difficult to delineate (for reviews, see Johnson, 2001; Johansson & Flanagan, 2009; Abraira & Ginty, 2013). The current consensus in the tactile literature is that the four main types of primary afferents are optimal for different functions. SA1s have high spatial acuity to skin indentation, which makes them the best candidate for coding tactile spatial information, such as stimulus position and curvature. SA2s are sensitive to skin stretch, which allows them to signal motion direction, velocity, hand position and finger conformation through the pattern of skin stretch. RAs are sensitive to low-frequency vibration (i.e. flutter), which endows them with remarkable efficiency in signalling sudden motion on the skin and providing feedback for slip and grip control. PCs are extremely sensitive to high- frequency vibration, which enables them to transmit distant tactile information through objects held in the hand (Macefield, 1998; Johnson, 2001; Abraira & Ginty, 2013). Traditionally, the four afferent types have often been viewed as playing largely non- overlapping roles in mediating these functions (Ochoa & Torebjörk, 1983; Bolanowski et al. 1988; Johnson et el. 2000; Johnson, 2001). This view is supported by neurophysiological and neuroelectrical evidence suggesting that segregation of signals from different afferent types extends to S1 (e.g. Mountcastle, 1956; Sur et al., 1981; Romo et al., 2000). However, more recent evidence has suggested that individual S1 Ph.D. Thesis – Li, L. McMaster University - Psychology 3 neurons receive convergent input from multiple afferent types, and that there may be a greater degree of functional interplay among the afferent types than traditionally believed (Pei et al., 2009; Saal & Bensmaia, 2014). Because the present thesis focuses on tactile spatial perception, here we will give a more specific overview of SA1 afferents, which are generally viewed as the main afferents responsible for coding spatial detail. SA1s can transmit a highly precise spatial image of tactile stimuli. Several physiological properties equip SA1s with this remarkable ability. First, SA1s have small, well-defined RFs with points of maximum firing (“hot spots”) within the fields. The hot spots correspond to individual branches of the afferent axon. When the stimulus is finer than the RF diameter (typically 2-3 mm on primate fingertips), an individual hot spot becomes dominant, allowing an individual SA1 afferent to resolve spatial detail as small as 0.5 mm (Phillips & Johnson, 1981; Phillips et al. 1992; Johnson, 2001). Second, SA1s innervate the skin with high density, about 1 afferent per mm2 in monkey fingertips (Johnson et al., 2000) and 0.7 afferent per mm2 in human fingertips (Johansson & Vallbo, 1979), and a single afferent can supply as many as 15 Merkel cells (Abraira & Ginty, 2013). Third, SA1s and Merkel cells are located close to the skin surface, in the basal layer of the epidermis (Halata et al., 2010); the shallow location facilitates their processing of tactile information on the skin surface. Fourth, SA1s are highly sensitive to points, edges, curvatures, and gaps, due to their selective sensitivity to local strain components on the skin (Phillips & Johnson, 1982; Sripati et al. 2006). Fifth, SA1s lack spontaneous firing. Sixth, SA1s are insensitive to skin displacement adjacent to their RFs or skin stretch (Johnson, 2001; Abraira & Ginty, 2013). These characteristics all contribute to SA1s’ high-fidelity coding of fine spatial information. The studies described in this thesis involved experimentation on human forearm skin, which belongs to the category of hairy skin. Microneurography studies have identified five classes of myelinated mechanoreceptive units in in human forearm skin (Vallbo et al., 1995; Olausson et al., 2000): two slowly-adapting types – SA1s and SA2s, and three fast-adapting types - hair units (also known as hair follicle afferents in the literature), field units, and Pacinian (PC) units. These mechanoreceptive units have also been identified in the hairy skin of a variety of mammals (for reviews, see Iggo & Andres, 1982; Zimmerman et al., 2014). Drawing from previous literature on mammalian hairy skin receptors, Vallbo et al. (1995) and Olausson et al. (2000) suggested that the end organs for SA1s, SA2s, hair units, and PC units in human forearm skin are Merkel cells, Ruffini endings, hair follicles, and Pacinian corpuscles, respectively; the end organs for field units are unclear. Less is known about hairy skin mechanoreceptive afferents than glabrous skin ones; however, studies have suggested that the characteristics of hairy skin mechanoreceptive afferents are somewhat similar to their counterparts in glabrous skin. For example, SA1, SA2, and PC afferents can be classified based on their RF characteristics in a similar fashion to those in glabrous skin: SA1 RFs are small, well-defined, comprising 2-4 highly sensitive spots that presumably correspond to clusters of Merkel cells innervated by that Ph.D. Thesis – Li, L. McMaster University - Psychology 4 afferent, and lacking spontaneous firing; they are capable of signaling spatiotemporal information (Vallbo et al., 1995; Olausson et al., 2000). SA2 RFs are highly sensitive to skin stretch and display spontaneous background discharge; they play important roles in proprioception and kinesthesia (Edin, 1992; Olausson et al., 2000). PCs are sensitive to high-frequency vibration and remote taps, have large RFs, and are located deep down the tissues near bones and joints (Merzenich & Harrington, 1969; Sahai et al., 2006). Apart from these similarities, hairy skin afferents also exhibit some characteristics that differ from those in glabrous skin. For example, the innervation density of SA1s is much lower in human forearm skin (~4 per 100 mm2; Vallbo et al. 1995) than in human fingertip (~100 per 100 mm2; Johnson, 2001). The low density presumably contributes to the fact that the forearm has a much poorer spatial resolution than the fingertip (Stevens & Choo, 1996), because receptor innervation density is one of the determinants for spatial acuity (Peters et al. 2009). Moreover, human forearm skin seems to lack the most abundant mechanoreceptive afferents observed in glabrous skin – RA afferents (Johnson et al. 2000) – but instead has two other types of fast-adapting afferents in addition to PCs: hair units and field units. Hair units exhibit some properties similar to those of RA afferents: they are fast adapting, with the peak of sensitivity to flutter around 20-50 Hz, and are efficient in detecting motion or air puffs on the skin surface. Unlike RA afferents, however, hair units in human forearm have large RFs, roughly an order of a magnitude larger than SA1 RFs in human forearm. Field unit RFs bear some resemblance to hair unit RFs, although the functional roles of field units are unclear (Vallbo et al., 1995). The large RFs of hair units and field units in human forearm skin presumably limit their contribution to the perception of fine spatial detail. The peripheral signals generated by these receptors and afferents then ascend the arm through the median and ulnar nerves, enter the spinal cord through the dorsal root ganglia, travel through the DCML pathway, and project onto S1 and S2. In S1, neurons encode tactile information within a spatial map. S1 of each hemisphere represents tactile sensations from the contralateral body parts in a topographically organized manner. In this “somatosensory homunculus”, adjacent neurons tend to have adjacent RFs on the body. The somatotopic mapping between peripheral RFs and S1 representations have been clearly demonstrated by neurophysiological experiments: tactile stimulation of a specific body part elicits neural response in the S1 region that is responsible for that body part; conversely, direct stimulation of the S1 region induces a tactile sensation localized in the corresponding body part, even though no actual tactile stimulus is delivered to that body part (for a review, see Serino & Haggard, 2010). S1 comprises Brodmann areas 3a, 3b, 1, and 2. Area 3a responds primarily to proprioceptive stimulation. Areas 3b and 1 respond primarily to cutaneous stimulation; they are responsible for coding many elementary features of tactile stimuli, such as location, orientation, edge, and motion direction. The majority of area 3b neurons have single-locus RFs, whereas area 1 neurons have larger RFs and more composite response properties. Area 2 responds to both cutaneous and proprioceptive stimulation; it integrates input from areas 3b and 1 to code more complex spatial features, such as contour curvature. The components of S1 project to S2, which further integrates tactile information and extracts higher-order, more Ph.D. Thesis – Li, L. McMaster University - Psychology 5 complex stimulus features. For example, S2 neurons exhibit tuning for curvature direction and play important roles in two-dimensional shape perception. S2 neurons typically have very large RFs; some of them span across the midline of the body and receive bilateral input. In general, as tactile signals travel from earlier cortical processing areas to downstream areas (S1 area 3b -> area 1 -> area 2 -> S2 -> further downstream), neural responses reflect increasingly complex and integrated stimulus features, and neural RFs encompass increasingly larger and more composite body regions. Similar to neurons in earlier processing areas, some neurons in downstream areas are tuned to some elementary stimulus features, but over much larger RFs. For example, similar to some neurons in S1 areas 3b and 1, some neurons in S1 area 2 and S2 are tuned to orientation, but over much larger skin regions, e.g. covering multiple digits or even both hands. For these higher- order neurons, orientation tuning tends to be consistent across their very large RFs, suggesting that orientation tuning becomes position-invariant (for reviews, see Iwamura, 1998; Bensmaia & Yau, 2011; Yau et al., 2016). Within this peripheral and central processing system, tactile spatial perception is mediated by the population response of groups of neurons, which is subject to the influence of sensory history and context. In other words, tactile spatial perception arises from the dynamic interplay among neuronal properties and sensory history. Sensory history mediates neuronal response properties and can substantially affect tactile spatial perception. 1.2 Adaptation and perceptual illusions The present thesis aims to investigate how tactile spatial perception is influenced by recent sensory history – specifically, sensory adaptation. Our sensory systems continuously adjust to the sensory environment. When we are exposed to a sustained sensory stimulus – for example, light, noise, a scent, or clothes that touch our skin, our sensory systems adjust neural responses to reserve energy and efficiently represent the environment. This phenomenon, known as adaptation, is ubiquitous in all sensory modalities. Adaptation influences perception; for example, the same stimulus often feels less intense after prolonged exposure because of adaptation. In tactile research, the term “adaptation” is usually used in two related but different contexts. The context that is used less is the general context as described in the previous paragraph: adaptation refers to the progressive changes in neural responses resulting from sustained stimulation, or the application of sustained stimulation to induce such changes. The other context, which is used much more frequently, is to classify mechanoreceptive afferents based on their response to sustained indentation, such as “slowly-adapting” (SA) or “rapidly-adapting” (RA) as described in Chapter 1.1. The same scheme has been used to classify somatosensory cortical neurons. For example, an S1 neuron that fires throughout a sustained stimulation on the skin, or receives input primarily from SA Ph.D. Thesis – Li, L. McMaster University - Psychology 6 afferents, is classified as an SA neuron. Even though response adaptation is one of the bases on which tactile receptors and, to a lesser extent, somatosensory cortical neurons are classified, the perceptual effects of tactile adaptation in the more general context have not been well studied. The present thesis drew inspiration from the rich literature on visual adaptation. Unlike in touch, the perceptual effects of adaptation in vision have been studied extensively. A topic of interest in the visual adaptation literature is adaptation-induced aftereffects, particularly perceptual illusions, because illusions provide clues as to how the visual system normally encodes and interprets stimuli. Adaptation is often used as a tool in perceptual studies to induce aftereffects or illusions, thereby probing neural selectivity and neural computations in information processing (for reviews, see Webster, 2012; Solomon & Kohn, 2014). A well-known adaptation-induced visual illusion is the tilt aftereffect (TAE) illusion: prolonged viewing of tilted lines causes subsequently viewed lines of a nearby orientation to appear tilted away from the adapted orientation (Gibson & Radner, 1937; Magnussen & Johnsen, 1986; Dragoi et al., 2000). The TAE has been studied extensively to probe the selectivity and functional organization of orientation-tuned neurons in the primary visual cortex (V1). The TAE is an example of a perceptual repulsion illusion: following focal adaptation, the subsequently viewed orientation appears to be “repelled” from the adapted orientation. In other words, focal adaptation induces a repulsive shift in the percept of the subsequent nearby orientation. Here, focal adaptation means to selectively adapt to a narrow, specific stimulus characteristic (in this example, a specific orientation); the degree of adaptation is likely to be graded as a function of distance from the adaptor value: neurons tuned to the adaptor orientation adapts the most, and neurons tuned to nearby orientations also adapts but to a lesser extent. The adaptation-induced repulsive shift has perceptual benefits: it effectively magnifies the perceived difference between the subsequent nearby orientation and the adaptor orientation, thus enhancing perceptual resolution around the adaptor orientation, as evidenced by reduced discrimination threshold (Schwartz et al., 2007). Adaptation- induced perceptual repulsion has been well documented for a wide variety of visual properties, including orientation, motion direction, position, curvature, size, contrast, spatial frequency, and even high-level features such as facial properties (for reviews, see Clifford et al., 2007; Kohn, 2007; Webster, 2012). The common existence of this phenomenon across categories of visual perception may point to a fundamental computational strategy in the visual system for information processing. It has been proposed that, following lengthy exposure to a sustained stimulus, the visual system adapts by recalibrating neural responses to match the new baseline, in order to preserve energy and increase sensitivity to changes; the repulsive aftereffect is a consequence and by-product of this recalibration. Despite having general functional benefits, the adaptation process can manifest as seemingly non-beneficial perceptual illusions under unusual physical conditions such as those experimentally manipulated (Stocker & Simoncelli, Ph.D. Thesis – Li, L. McMaster University - Psychology 7 2006; Kohn, 2007; Seriès et al., 2009; Fischer & Whitney, 2014). In light of the visual literature, an interesting question arises: Does adaptation induce an analogous repulsive effect in tactile perception? Touch and vision share many similarities in their perceptual goals and functional organizations; for example, both systems need to extract information from two-dimensional receptor sheets, both systems have analogous sensory channels optimized for coding certain spatial or temporal features, and both systems exhibit neural tuning to location and orientation. Although the physiological substrates of sensory transduction are by necessity different in these two systems, it has long been speculated that touch and vision have similar functional mechanisms for coding and representing information, especially spatial information (for a review, see Hsiao, 1998). Therefore, it is plausible that focal adaptation causes the tactile system to undergo similar changes as in the visual system, which leads to a perceptual repulsion for subsequent stimuli whose properties (e.g. orientation, position, frequency) are close to the adaptor value. A manifestation of the repulsive effect would be that discriminability around the adaptor value is enhanced following adaptation. Indeed, both the repulsive effect and enhanced discriminability following adaptation have been observed in tactile perception. Tactile psychophysical studies have reported that focal adaptation leads to perceptual repulsion aftereffects in motion direction (McIntyre et al., 2016a), speed (McIntyre et al., 2016a, 2016b), orientation (Silver, 1969), and distance between two simultaneous stimuli (Day & Singer, 1964; Calzolari et al. 2017), as well as improving discrimination performance in vibrotactile frequency (Goble & Hollins, 1994; Tommerdahl et al., 2005; Tannan et al., 2007), amplitude (Goble & Hollins, 1993; Delemos & Hollins, 1996), and spatial localization (Tannan et al., 2006). Of these effects, the impact of adaptation on tactile spatial perception (e.g. position, distance) is the least studied and the most inconclusive. The very few existing studies on this topic have yielded somewhat contradictory results. For example, an early study on human forearm skin reported that adaptation altered the perceived separation between parallel bars placed on adjacent skin areas in a direction consistent with perceptual repulsion (Day & Singer, 1964). A follow-up study suggested, however, that the observed effects may not be adaptation-induced aftereffects, but rather estimation artifacts induced by the particular sets of comparison stimuli to which the participants were exposed (Gilbert, 1967). In the present thesis, we revisited the question of whether adaptation induces spatial repulsion in touch. Specifically, we examined whether focal vibratory adaptation on the forearm induces a spatial repulsion illusion affecting the perceived distance between two points of contact straddling the adapted region. Punctate point-stimuli are commonly used to measure tactile spatial processing. Locating point-stimuli and determining the distance between them are the basis of many tactile spatial tasks. The present thesis provides one of the first studies investigating adaptation effects on tactile spatial perception, a topic that has rarely been documented in the literature. Ph.D. Thesis – Li, L. McMaster University - Psychology 8 1.3 Modeling perception as probabilistic inference How do nervous systems transform raw sensory information into perception? Neurophysiological and psychophysical studies, among other empirical research, have helped to shed light on this question by revealing bits and pieces of the puzzle. However, there lacks a unifying theory on how nervous systems code, represent, and store sensory information, and how perception arises from these procedures. A promising complement to the empirical research is computational modeling, which aims to tackle the overarching coding principles that theoretically govern perception. The present thesis implements a computational framework known as Bayesian inference, which views perception as a probabilistic inference. The idea of perception as a problem of inference can be dated back to Aristotle. In the 19th century, Hermann von Helmholtz systematically developed the concept of perception as an unconscious inference; he used the visual illusion of the sun rotating around the earth as an example to illustrate this inference. The task of perception is to infer the properties of the external environment from the patterns of sensorineural responses. A fundamental challenge faced by perception is the inherent uncertainty at every stage of processing. Sources of uncertainty include ambiguous stimuli, low receptor density, stochasticity in neural firing, and the multitude of hypothetical scenarios that are consistent with the available sensory data. The inherent uncertainty and noise in perception is best described in probabilistic terms. Over the past several decades, researchers have rigorously applied concepts from probability theory and information theory to investigate problems in perception and other neuroscience topics. One of the powerful probabilistic frameworks they have applied is Bayesian inference (Rao et al., 2002; Knill & Pouget, 2004). The present thesis implements computational modeling with Bayesian inference. The Bayesian models treat perception as a statistical inference consisting of two information- processing stages: encoding and decoding. Encoding is the forward-processing, data- generative stage, in which stimulus properties are transformed into sensory data (e.g. firing rates) in the form of probabilistic measurements (i.e. likelihood probabilities). Decoding deals with the inverse problem; a Bayesian observer interprets the noisy sensory data in light of sensory experience or expectation (i.e. prior probabilities) to provide probabilistic estimates (i.e. posterior probabilities) for the stimulus properties (Knill & Pouget, 2004; Goldreich, 2007). Thus, Bayesian inference allows the model observer to quantify uncertainty in different stages of perception in a unified and well- controlled manner. With Bayesian modeling, we aimed to better understand the perceptual effects that we empirically observed. The Bayesian model yielded psychometric functions that allowed us to quantitatively compare the model performance with the psychophysically measured human performance. Moreover, it allowed us to explore the factors that plausibly contributed to the empirical observations in a simulated environment, where we could Ph.D. Thesis – Li, L. McMaster University - Psychology 9 specifically define parameters, directly manipulate constraints and information available, and precisely measure responses. Our goal was not only to replicate human performance – which we did – but also to predict human performance given different constraints, and to shed light on the possible neural response properties and computations that underlie tactile spatial perception and adaption-induced repulsion illusions. 1.4 Overview of studies The present thesis investigated the effects of adaptation on tactile spatial perception, using a combination of psychophysical experimentation and computational modeling. With psychophysical testing in human participants, we showed that tactile spatial perception is subject to an adaptation-induced repulsion illusion that expands the perceived distance between points on the skin (Chapter 2). This illusion, however, was not observed when the intervening skin between points was desensitized with topical anesthesia instead of vibrotactile adaptation (Chapter 3). With Bayesian perceptual models, we showed that the repulsion illusion empirically observed was consistent with the hypothesis that the brain decodes the tactile spatial input without awareness of the adaptation state in the nervous system (Chapter 4). In Chapter 2, we examined the effects of vibrotactile adaptation on two-point distance perception. In a series of experiments involving a two-interval forced-choice (2IFC) task, participants compared the perceived separation between two point-stimuli applied on the forearms successively. Separation distance was constant on one arm (the reference) and varied on the other arm (the comparison). Experiment 1 applied repeated baseline measurements, and verified that participants’ distance perception was unbiased across arms and stable across experimental blocks. Experiment 2 implemented a monofilament- detection task, and showed that vibration of the skin between the two stimulus points on the reference arm focally reduced tactile sensitivity, verifying the efficacy of the vibrotactile protocol in inducing adaptation. Experiment 3 repeated the distance- comparison task in Experiment 1 with the adaptation protocol from Experiment 2, and showed that adaptation significantly increased the perceived distance between the reference points, causing a repulsion illusion. The results are consistent with findings in the visual and auditory perception literature that reported repulsion illusions following focal adaptation. In Chapter 3, we conducted the battery of tests from Chapter 2 in a different group of participants, but applied a topical anesthetic (a mixture of lidocaine and prilocaine) instead of vibration to the intervening skin between the reference points. Anesthesia focally reduced tactile sensitivity but caused little to no increase in perceived two-point distance. We discussed possible explanations for the discrepancy between the adaptation and anesthesia results. A possibility is that mere desensitization of peripheral receptors is Ph.D. Thesis – Li, L. McMaster University - Psychology 10 not sufficient to cause the repulsion illusion observed in Chapter 2, and that adaptation in the central nervous system is also required. In Chapter 4, we implemented Bayesian perceptual models to investigate adaptation effects on tactile spatial perception on a theoretical and computational basis. The model has two major components: a generative model (the encoder) and a Bayesian decoder. The generative model simulated somatosensory neural firing patterns evoked by point- stimuli. It incorporated response properties of somatosensory cortical neurons, including the spacing and size of their receptive fields, firing rate variability, and adaptation state. The Bayesian decoder interpreted the simulated neural data from the generative model to perform 2IFC tasks (two-point distance comparison, monofilament detection). With specific sub-optimal constraints, such as sparse receptive fields and Poisson firing noise, the Bayesian observer performed quantitatively similarly to human participants. It exhibited a repulsion illusion following adaptation in the two-point distance comparison task, which was comparable to the repulsion illusion we empirically observed in Chapter 2. In general, this illusion emerged when the decoder was unaware of the adaptation in the encoding stage. We speculated on the plausibility of this assumption, as well as other implications from the model performance, in the context of human tactile perception and its underlying neural computations. Taken together, this thesis provides one of the first combined psychophysical and computational studies on the effects of adaptation on tactile spatial perception. Its main findings are consistent with the adaptation-induced repulsion illusions reported in vision and audition, suggesting that tactile perception shares common processing features with visual and auditory perception. It sheds light on possible mechanisms and functional organizations underlying dynamic tactile spatial processing as the somatosensory system adjusts to the external environment. 1.5 References Abraira, V., & Ginty, D. (2013). The sensory neurons of touch. Neuron, 79(4), 618–639. https://doi.org/10.1016/j.neuron.2013.07.051 Bolanowski, S. J., Gescheider, G. a, Verrillo, R. T., & Checkosky, C. M. (1988). Four channels mediate the mechanical aspects of touch. The Journal of the Acoustical Society of America, 84(5), 1680–1694. https://doi.org/10.1121/1.397184 Calzolari, E., Azanon, E., Danvers, M., Vallar, G., & Longo, M. R. (2017). Adaptation aftereffects reveal that tactile distance is a basic somatosensory feature. Proc Natl Acad Sci USA, 114(17), 4555–4560. Clifford, C. W. G., Webster, M. A., Stanley, G. B., Stocker, A. A., Kohn, A., Sharpee, T. O., & Schwartz, O. (2007). Visual adaptation: Neural, psychological and computational aspects. Vision Research, 47(25), 3125–3131. Ph.D. Thesis – Li, L. McMaster University - Psychology 11 https://doi.org/10.1016/j.visres.2007.08.023 Day, R. H., & Singer, G. (1964). A tactile spatial aftereffect. Australian Journal of Psychology, 16(I), 33–37. Delemos, K. a, & Hollins, M. (1996). Adaptation-induced enhancement of vibrotactile amplitude discrimination: the role of adapting frequency. The Journal of the Acoustical Society of America, 99(1), 508–16. https://doi.org/10.1121/1.414509 Dragoi, V., Sharma, J., & Sur, M. (2000). Adaptation-induced plasticity of orientation tuning in adult visual cortex. Neuron, 28(1), 287–298. https://doi.org/10.1016/S0896-6273(00)00103-3 Edin, B. B. (1992). Quantitative analysis of static strain sensitivity in human mechanoreceptors from hairy skin. Journal of Neurophysiology, 67(5), 1105–1113. https://doi.org/10.1152/jn.00628.2004. Fischer, J., & Whitney, D. (2014). Serial dependence in visual perception. Nature Publishing Group, 17(5), 738–743. https://doi.org/10.1038/nn.3689 Fulkerson, M. (2016). “Touch”. The Stanford Encyclopedia of Philosophy. (Spring 201). Stanford University. Retrieved from https://plato.stanford.edu/entries/touch/ Gibson, J. J., & Radner, M. (1937). Adaptation, After-Effect and Constrast in the Perception of Tilted Lines. Journal of Experimental Psychology, 20(5), 453–467. https://doi.org/10.1037/h0059826 Gilbert, A. J. (1967). Tactile spatial aftereffect or adaptation level? Journal of Experimental Psychology, 73(3), 450–455. Goble, A. K., & Hollins, M. (1993). Vibrotactile adaptation enhances amplitude discrimination. The Journal of the Acoustical Society of America, 93(1), 771–780. https://doi.org/10.1121/1.410314 Goble, A. K., & Hollins, M. (1994). Vibrotactile adaptation enhances frequency discrimination. The Journal of the Acoustical Society of America, 96(2), 771–780. https://doi.org/10.1121/1.410314 Goldreich, D. (2007). A Bayesian perceptual model replicates the cutaneous rabbit and other tactile spatiotemporal illusions. PLoS ONE, 2(3). https://doi.org/10.1371/journal.pone.0000333 Halata, Z., Baumann, K. I., & Grim, M. (2010). Merkel Cells. The Senses: A Comprehensive Reference, 6, 33–38. https://doi.org/10.1016/B978-012370880- 9.00341-8 Hsiao, S. S. (1998). Similarities between touch and vision. In J. W. Morley (Ed.), Neural computation (pp. 131–165). Amsterdam ; New York: Elsevier. Iggo, A., & Andres, K. H. (1982). Morphology of Cutaneous Receptors. Ann. Rev. Neurosci., (5), 1–31. Iwamura, Y. (1998). Representation of tactile functions in the somatosensory cortex. In J. W. Morley (Ed.), Neural Aspects of Tactile Sensation (pp. 195–238). Elsevier B.V. Johansson, R. S., & Flanagan, J. R. (2009). Coding and use of tactile signals from the fingertips in object manipulation tasks. Nature Reviews Neuroscience, 10(5), 345– 359. https://doi.org/10.1038/nrn2621 Johansson, R., & Vallbo, A. (1979). Tactile sensibility in the human hand: relative and absolute densities of four types of mechanoreceptive units in glabrous skin. The Ph.D. Thesis – Li, L. McMaster University - Psychology 12 Journal of Physiology, (286), 283–300. https://doi.org/10.1113/jphysiol.1979.sp012619 Johnson, K. O. (2001). The Roles and Functions of cutaneous mechanoreceptors. Current Opinion in Neurobiology, 11, 455–461. Retrieved from papers3://publication/uuid/5F042AE7-BBCB-4A6C-B1BA-B6480F271B1B Johnson, K. O., & Hsiao, S. S. (1992). Neural mechanisms of tactile form and texture perception. Ann. Rev. Neurosci., 15, 227–250. https://doi.org/10.1016/j.cub.2014.02.049 Johnson, K. O., Yoshioka, T., & Vega-Bermudez, F. (2000). Tactile functions of mechanoreceptive afferents innervating the hand. Journal of Clinical Neurophysiology : Official Publication of the American Electroencephalographic Society, 17(6), 539–558. https://doi.org/10.1097/00004691-200011000-00002 Knill, D. C., & Pouget, A. (2004). The Bayesian brain: The role of uncertainty in neural coding and computation. Trends in Neurosciences, 27(12), 712–719. https://doi.org/10.1016/j.tins.2004.10.007 Kohn, A. (2007). Visual Adaptation: Physiology, Mechanisms, and Functional Benefits. Journal of Neurophysiology, 97(5), 3155–3164. https://doi.org/10.1152/jn.00086.2007 Macefield, V. G. (1998). The signalling touch, movement and manipulation by mechanoreceptors in human skin. In J. W. Morley (Ed.), Neural Aspects of Tactile Sensation (pp. 89–130). Amsterdam: Elsevier. Magnussen, S., & Johnsen, T. (1986). Temporal aspects of spatial adaptation. A study of the tilt aftereffect. Vision Research, 26(4), 661–672. https://doi.org/10.1016/0042- 6989(86)90014-3 McGlone, F., & Reilly, D. (2010). The cutaneous sensory system. Neuroscience and Biobehavioral Reviews, 34(2), 148–159. https://doi.org/10.1016/j.neubiorev.2009.08.004 McIntyre, S., Birznieks, I., Vickery, R. M., Holcombe, A. O., & Seizova-Cajic, T. (2016). The tactile motion aftereffect suggests an intensive code for speed in neurons sensitive to both speed and direction of motion. Journal of Neurophysiology, 115(3), 1703–1712. https://doi.org/10.1152/jn.00460.2015 McIntyre, S., Seizova-Cajic, T., & Holcombe, A. O. (2016). The tactile speed aftereffect depends on the speed of adapting motion across the skin rather than other spatiotemporal features. Journal of Neurophysiology, 115(3), 1112–1121. https://doi.org/10.1152/jn.00821.2014 Morley, J. W. (Ed.). (1998). Neural aspects in tactile sensation / edited by J.W. Morley. Amsterdam ; New York: Elsevier. Retrieved from http://www.loc.gov/catdir/enhancements/fy0602/98209962-t.html Mountcastle, V. (1957). Modality and topographic properties neurons of cat’s somatic sensory. Journal of Neurophysiology, 20(4), 408–434. Ochoa, J., & Torebjörk, E. (1983). Sensations evoked by intraneural microstimulation of single mechanoreceptor units innervating the human hand. The Journal of Physiology, 342, 633–54. https://doi.org/10.1016/0304-3959(85)90304-5 Olausson, H., Wessberg, J., & Kakuda, N. (2000). Tactile directional sensibility: Ph.D. Thesis – Li, L. McMaster University - Psychology 13 Peripheral neural mechanisms in man. Brain Research, 866(1–2), 178–187. https://doi.org/10.1016/S0006-8993(00)02278-2 Pei, Y.-C., Denchev, P. V, Hsiao, S. S., Craig, J. C., & Bensmaia, S. J. (2009). Convergence of submodality-specific input onto neurons in primary somatosensory cortex. Journal of Neurophysiology, 102(3), 1843–1853. https://doi.org/10.1152/jn.00235.2009 Peters, R. M., Hackeman, E., & Goldreich, D. (2009). Diminutive Digits Discern Delicate Details: Fingertip Size and the Sex Difference in Tactile Spatial Acuity. Journal of Neuroscience, 29(50), 15756–15761. https://doi.org/10.1523/JNEUROSCI.3684- 09.2009 Phillips, J. R., & Johnson, K. O. (1981). Tactile spatial resolution. III. A continuum mechanics model of skin predicting mechanoreceptor responses to bars, edges, and gratings. Journal of Neurophysiology, 46(6), 1204–1225. https://doi.org/10.1152/jn.00955.2003 Romo, R., Hernández, a, Zainos, a, Brody, C. D., & Lemus, L. (2000). Sensing without touching: psychophysical performance based on cortical microstimulation. Neuron, 26(1), 273–278. https://doi.org/10.1016/S0896-6273(00)81156-3 Rustioni, A., Hayes, N. L., & O’Neill, S. (1979). Dorsal column nuclei and ascending spinal afferents in macaques. Brain : A Journal of Neurology, 102(1), 95–125. https://doi.org/https://doi.org/10.1093/brain/102.1.95 Saal, H. P., & Bensmaia, S. J. (2014). Touch is a team effort: Interplay of submodalities in cutaneous sensibility. Trends in Neurosciences, 37(12), 689–697. https://doi.org/10.1016/j.tins.2014.08.012 Schwartz, O., Hsu, A., & Dayan, P. (2007). Space and time in visual context. Nature Reviews. Neuroscience, 8(7), 522–535. https://doi.org/10.1038/nrn2155 Seriès, P., Stocker, A. A., & Simoncelli, E. P. (2009). Is the homunculus “aware” of sensory adaptation? Neural Computation, 21(12), 3271–304. https://doi.org/10.1162/neco.2009.09-08-869 Serino, A., & Haggard, P. (2010). Touch and the body. Neuroscience and Biobehavioral Reviews, 34(2), 224–236. https://doi.org/10.1016/j.neubiorev.2009.04.004 Silver, R. J. (1969). Tilt after-effects in touch. Brandeis University. Solomon, S. G., & Kohn, A. (2014). Moving sensory adaptation beyond suppressive effects in single neurons. Current Biology, 24(20), R1012–R1022. https://doi.org/10.1016/j.cub.2014.09.001 Sripati, A. P., Bensmaia, S. J., & Johnson, K. O. (2006). A Continuum Mechanical Model of Mechanoreceptive Afferent Responses to Indented Spatial Patterns. Journal of Neurophysiology, 95(6), 3852–3864. https://doi.org/10.1152/jn.01240.2005 Stevens, J. C., & Choo, K. K. (1996). Spatial Acuity of the Body Surface over the Life Span. Somatosensory & Motor Research, 13(2), 153–166. https://doi.org/10.3109/08990229609051403 Stocker, A. a, & Simoncelli, E. P. (2006). Noise characteristics and prior expectations in human visual speed perception. Nature Neuroscience, 9(4), 578–585. https://doi.org/10.1038/nn1669 Sur, M., Wall, J. T., & Kaas, J. H. (1981). Modular segregation of functional cell classes Ph.D. Thesis – Li, L. McMaster University - Psychology 14 within the postcentral somatosensory cortex of monkeys. Science (New York, N.Y.), 212(4498), 1059–1061. https://doi.org/10.1126/science.7233199 Sur, M., Wall, J. T., & Kaas, J. H. (1984). Modular distribution of neurons with slowly adapting and rapidly adapting responses in area 3b of somatosensory cortex in monkeys. Journal of Neurophysiology, 51(4), 724–744. Tannan, V., Simons, S., Dennis, R. G., & Tommerdahl, M. (2007). Effects of adaptation on the capacity to differentiate simultaneously delivered dual-site vibrotactile stimuli. Brain Research, 1186(1), 164–170. https://doi.org/10.1016/j.brainres.2007.10.024 Tannan, V., Whitsel, B. L., & Tommerdahl, M. A. (2006). Vibrotactile adaptation enhances spatial localization. Brain Research, 1102(1), 109–116. https://doi.org/10.1016/j.brainres.2006.05.037 Tommerdahl, M., Hester, K. D., Felix, E. R., Hollins, M., Favorov, O. V., Quibrera, P. M., & Whitsel, B. L. (2005). Human vibrotactile frequency discriminative capacity after adaptation to 25 Hz or 200 Hz stimulation. Brain Research, 1057(1–2), 1–9. https://doi.org/10.1016/j.brainres.2005.04.031 Vallbo, A. B., Olausson, H., Wessberg, J., & Kakuda, N. (1995). Receptive field characteristics of tactile units with myelinated afferents in hairy skin of human subjects. The Journal of Physiology, 483, 783–95. https://doi.org/10.1113/jphysiol.1995.sp020622 Webster, M. (2012). Evolving concepts of sensory adaptation. F1000 Biology Reports, 4(November), 1–7. https://doi.org/10.3410/B4-21 Yau, J. M., Kim, S. S., Thakur, P. H., & Bensmaia, S. J. (2016). Feeling form: the neural basis of haptic shape perception. Journal of Neurophysiology, 115(2), 631–642. https://doi.org/10.1152/jn.00598.2015 Zimmerman, A., Bai, L., & Ginty, D. D. (2014). The gentle touch receptors of mammalian skin. Science (New York, N.Y.), 346(6212), 950–4. https://doi.org/10.1126/science.1254229 Ph.D. Thesis – Li, L. McMaster University - Psychology 15 CHAPTER 2 ADAPTATION 2.1 Preface The perceptual effects of sensory adaptation are well documented in vision and audition, but have been much less studied in touch. In this chapter, we investigated the effects of adaptation on tactile spatial perception. With psychophysical testing involving two- interval forced-choice (2IFC) tasks in human participants, we measured tactile sensitivity and the perceived distance between point-stimuli on the forearm skin with and without vibratory adaptation. We found that adaptation significantly reduced tactile sensitivity and induced a repulsion illusion in tactile spatial perception that expands the perceived distance between points on the skin. This study is one of the first to provide evidence for adaptation-induced spatial repulsion illusions in touch. The results are consistent with adaptation-induced repulsion illusions reported in vision and audition, and may point to common processing features and computational principles across the sensory modalities. ORIGINAL RESEARCH published: 28 June 2017 doi: 10.3389/fnhum.2017.00331 An Adaptation-Induced Repulsion Illusion in Tactile Spatial Perception Lux Li 1, Arielle Chan1, Shah M. Iqbal1 and Daniel Goldreich1,2* 1Department of Psychology, Neuroscience and Behaviour, McMaster University, Hamilton, ON, Canada, 2McMaster Integrative Neuroscience Discovery and Study, McMaster University, Hamilton, ON, Canada Edited by: Christoph Braun, University of Tübingen, Germany Reviewed by: Alessandro Farne, Institut National de la Santé et de la Recherche Médicale (INSERM), France Matthew R. Longo, Birkbeck University of London, United Kingdom *Correspondence: Daniel Goldreich goldrd@mcmaster.ca Received: 09 February 2017 Accepted: 08 June 2017 Published: 28 June 2017 Citation: Li L, Chan A, Iqbal SM and Goldreich D (2017) An Adaptation-Induced Repulsion Illusion in Tactile Spatial Perception. Front. Hum. Neurosci. 11:331. doi: 10.3389/fnhum.2017.00331 Following focal sensory adaptation, the perceived separation between visual stimuli that straddle the adapted region is often exaggerated. For instance, in the tilt aftereffect illusion, adaptation to tilted lines causes subsequently viewed lines with nearby orientations to be perceptually repelled from the adapted orientation. Repulsion illusions in the nonvisual senses have been less studied. Here, we investigated whether adaptation induces a repulsion illusion in tactile spatial perception. In a two-interval forced-choice task, participants compared the perceived separation between two point- stimuli applied on the forearms successively. Separation distance was constant on one arm (the reference) and varied on the other arm (the comparison). In Experiment 1, we took three consecutive baseline measurements, verifying that in the absence of manipulation, participants’ distance perception was unbiased across arms and stable across experimental blocks. In Experiment 2, we vibrated a region of skin on the reference arm, verifying that this focally reduced tactile sensitivity, as indicated by elevated monofilament detection thresholds. In Experiment 3, we applied vibration between the two reference points in our distance perception protocol and discovered that this caused an illusory increase in the separation between the points. We conclude that focal adaptation induces a repulsion aftereffect illusion in tactile spatial perception. The illusion provides clues as to how the tactile system represents spatial information. The analogous repulsion aftereffects caused by adaptation in different stimulus domains and sensory systems may point to fundamentally similar strategies for dynamic sensory coding. Keywords: somatosensory, psychophysics, sensory adaptation, perceptual inference, tactile illusion, two-point perception, human, aftereffect INTRODUCTION Prolonged exposure to stimulation causes a reduction in neuronal firing rate. For reasons that have yet to be elucidated, this phenomenon, adaptation, is ubiquitous in neural sensory systems (Wark et al., 2007; Sato and Aihara, 2011). Adaptation may have several beneficial consequences: it may support perceptual constancy, increase the salience of novel stimuli, improve discrimination and improve coding efficiency (for review see Webster, 2012). A seemingly non-beneficial consequence of focal adaptation is that it produces illusions. For instance, following focal adaptation, the perceived separation between stimuli that straddle the adapted region is often exaggerated. A well-known example of this is the visual tilt after effect Frontiers in Human Neuroscience | www.frontiersin.org 1 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org/Human_Neuroscience/editorialboard http://www.frontiersin.org/Human_Neuroscience/editorialboard https://doi.org/10.3389/fnhum.2017.00331 http://crossmark.crossref.org/dialog/?doi=10.3389/fnhum.2017.00331&domain=pdf&date_stamp=2017-06-28 http://journal.frontiersin.org/article/10.3389/fnhum.2017.00331/abstract http://journal.frontiersin.org/article/10.3389/fnhum.2017.00331/abstract http://loop.frontiersin.org/people/398969/overview http://loop.frontiersin.org/people/415333/overview http://loop.frontiersin.org/people/451420/overview http://loop.frontiersin.org/people/74818/overview https://creativecommons.org/licenses/by/4.0/ mailto:goldrd@mcmaster.ca https://doi.org/10.3389/fnhum.2017.00331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion illusion: adaptation to tilted lines causes subsequently viewed lines with nearby orientations to appear tilted away, i.e., repelled, from the adapted orientation (Gibson and Radner, 1937; Magnussen and Johnsen, 1986; Dragoi et al., 2000, 2001; He and MacLeod, 2001). In vision, adaptation-induced repulsion illusions have been reported to affect perception of a wide variety of stimulus features, including luminance, contrast, spatial frequency, temporal frequency, color, contour, shape, size, orientation, motion direction, contingent visual properties (e.g., color and orientation, as in the McCollough effect) and high-level features such as the gender, ethnicity and emotion of faces (for reviews, see Clifford et al., 2007; Kohn, 2007; Webster, 2012). Adaptation- induced repulsive aftereffects have also been reported in auditory perception and audio-visual perception, including aftereffects in sound localization (Thurlow and Jack, 1973; Kashino and Nishida, 1998; Carlile et al., 2001), duration (Walker et al., 1981; Heron et al., 2012), loudness (Kitagawa and Ichihara, 2002), and high-level auditory perception such as action sounds (Barraclough et al., 2017). The present study concerns a particular type of adaptation- induced repulsion illusion, spatial repulsion, in which the positions of stimuli are perceptually repelled away from an adapted area. Spatial repulsion illusions have been well documented in vision (Clifford et al., 2007; Kohn, 2007; Schwartz et al., 2007) and to a lesser extent in audition (Kashino and Nishida, 1998; Carlile et al., 2001) but have rarely been reported in touch. An early tactile study reported that prolonged static pressure on the forearm altered the perceived separation between parallel bars placed on adjacent skin areas in a direction consistent with perceptual repulsion (Day and Singer, 1964). A follow-up study suggested, however, that the observed effects may not have been aftereffects but rather perceptual recalibrations induced by the particular sets of comparison stimuli to which the participants were exposed (Gilbert, 1967). Here, we revisited the question of whether adaptation-induced spatial repulsion occurs in touch. Specifically, we investigated whether focal vibratory adaptation on the forearm induces a spatial repulsion illusion affecting the perceived distance between two points of contact straddling the adapted region. We hypothesized that adaptation of the mechanoreceptors in the intervening skin would decrease the overlap between the neuronal population responses elicited by the two points. Consequently, the brain would infer a greater distance between the points: a repulsion illusion. MATERIALS AND METHODS Participants Sixty-nine participants were recruited from the McMaster University community. By self-report, all participants were free of conditions that are known to impair tactile sensitivity (e.g., calluses, scars, or injuries on tested skin areas, carpel tunnel syndrome, diabetes) or perceptual processing (e.g., neurological disorders, attention deficit disorders, dyslexia). All participants had normal or corrected-to-normal vision. Of the 69 recruits, 60 passed the perceptual qualification criteria (see below). Of the 60 qualified participants, 20 took part in Experiment 1 (13 women, 7 men; 17 right-handed, 2 left-handed, 1 ambidextrous; aged 18.7–30.5 years, median age 20.7 years), 20 in Experiment 2A (13 women, 7 men; 19 right-handed, 1 left-handed; aged 18.5–22.6 years, median age 19.9 years), and 20 in Experiments 2B and 3 (12 women, 8 men; all right-handed; aged 19.1–28.8 years, median age 20.8 years). Handedness was assessed by a modified Edinburgh Handedness Inventory (Oldfield, 1971). Participants provided signed informed consent and received monetary compensation and/or course credits for their participation. This study was carried out in accordance with the recommendations of the McMaster Research Ethics Board. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the McMaster Research Ethics Board. EXPERIMENT 1 Experiment 1 assessed whether the baseline perception of two-point distance was stable across experimental blocks and unbiased across arms. We tested participants on a two-interval forced-choice (2IFC) two-point distance comparison task to measure their baseline two-point distance perception. Preparation and Skin Sites Tested The participant sat in front of a table with the experimental apparatus concealed by an opaque black curtain. The participant’s forearms, inserted under the curtain, rested comfortably on a padded surface, with the wrists (palm side up) resting stably on concave foam supports. To assist the experimenters in positioning the stimuli, the participant’s forearms were demarcated with a fine-tipped pen. A pair of small dots 30 mm apart was drawn on each volar forearm to guide the application of the two-point test stimuli. On each arm, the dots were symmetrical about the midpoint between the wrist and the elbow, aligned with the proximal-distal axis of the forearm, and slightly laterally offset from midline (Figure 1A, left). The slightly lateral-to-midline skin surface was parallel to the ground when participants rested their forearms in a supine position as they naturally tended to rotate the forearms slightly inward when relaxed; the choice of this skin surface thereby facilitated the application of the test stimuli perpendicularly to participants’ forearms. Psychophysical Procedure A two-point stimulus was applied onto the participant’s volar forearm with the two points simultaneously indenting the skin. Approximately 1 s later, another two-point stimulus was applied to the other volar forearm. The participant compared the distance between the first pair of points with the distance between the second pair of points, and reported which distance felt greater (Figure 1A). The participants verbalized their answers by saying ‘‘first’’ or ‘‘second’’, and the experimenter recorded the answers into a computer by pressing one of two response keys. The two-point distance was fixed at 30 mm on the right forearm (the reference) and variable Frontiers in Human Neuroscience | www.frontiersin.org 2 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion FIGURE 1 | Experimental protocols and expected results. (A) Experiment 1: baseline distance comparison. Left: participants compared the perceived distance between two-point stimuli applied on the forearms successively. On the right arm, the points were separated by a fixed reference distance (R = 30 mm); on the left arm, the points were separated by a variable comparison distance (C = 6, 12, 18, 24, 30, 36, 42, 48, or 54 mm). Right: expected psychometric function. Horizontal axis: comparison distance, C. Vertical axis: proportion of trials in which the participant responds that C is greater than R, Presp(C > R). The point of subjective equality (PSE; vertical dashed line) is the value of C for which Presp(C > R) = 0.5 (horizontal dashed line); the expected PSE is equal to R. (B) Experiment 2: effect of adaptation on tactile sensitivity. Left: participants reported in which of two intervals they felt a monofilament stimulus on the right forearm. Circle: site of vibratory stimulus. Experiment 2A measured reduction in tactile sensitivity at the center of the vibration site under different adaptation protocols. Experiment 2B measured reduction in tactile sensitivity as a function of distance from the center of vibration. Right: expected results from Experiments 2A (top) and 2B (bottom). Monofilaments applied in a 2-down 1-up staircase procedure. Black, no adaptation (NA); blue, 40 s adaptation with no top-ups (A0); magenta, 40 s adaptation with 3 s top-ups (A3); yellow, 40 s adaptation with 7 s top-ups (A7). (C) Experiment 3: distance comparison, as in Experiment 1, but with and without adaptation. Left: circle: skin site that received vibratory adaptation. Right: expected psychometric functions. Horizontal and vertical axes as in (A). Black, NA; magenta, 40 s adaptation with 3 s top-ups (A3). A rightward shift upon adaptation (arrow) indicates increased perceived distance between points straddling the adapted skin site. from 6 mm to 54 mm in increments of 6 mm on the left forearm (the comparison; nine comparison distances in total). The application order of the reference and comparison points was counter balanced across participants: half of the participants received the reference points first and comparison second in all trials, and the other half of the participants received the comparison points first and reference second in all trials. Each participant completed a practice block followed by three identical testing blocks. The practice block consisted of 16 trials with auditory feedback to indicate whether the response was correct (two trials were presented for each of the eight comparison distances not equal to the reference distance of 30 mm). Each testing block consisted of 90 trials without feedback, 10 trials at each of the nine comparison distances, randomly sampled without replacement. A custom computer program (LabVIEW 2011 for Macintosh, National Instruments) instructed the experimenter as to which comparison distance to apply. The participant took a 5 min break after the practice and a 20 min break between testing blocks. During each testing block, the participant took a 1-min break upon completing each quarter of the 90 trials (i.e., after completing trials 22, 45, 67). Force-Controlled Two-Point Stimuli A custom-made lever system (Figure 2) was used to apply two-point stimuli in alignment with the proximal-distal axis of the forearm, and with force control. Each two-point applicator was made of two plastic pins attached to one face of the shaft of a wood pencil of hexagonal cross-section. The uniform size and weight of the pencils facilitated force control of the test stimuli, and the hexagonal cross-section helped align the two pins. The heights by which the pins protruded from the pencils were carefully adjusted such that they were equal for a given two-point applicator and across all applicators. The stimulus surfaces were spherical pinheads of diameter 1.5 mm. Separation distances between the centers of the pinheads were 6, 12, 18, 24, 30, 36, 42, 48 and 54 mm. The lever system consisted of two acetal plastic arms attached via a metal rod that passed through a ball bearing. The metal rod rotated with little friction, allowing the arms to swivel smoothly. A magnet was attached to the end of each arm, and two magnets were attached to each applicator. The applicator could be easily attached to and removed from the swivel arms via the magnets, which allowed the experimenter to quickly change the applicator from trial to trial. To apply a test stimulus, the experimenter first attached the applicator to the swivel arms. Supporting the swivel arms with both hands from below, the experimenter gently lowered the swivel arms such that the two pinheads contacted the forearm simultaneously and perpendicular to the skin surface. The pinheads contacted the skin with a total force determined by the combined weight of the swivel arms, which measured 80–82 g when the pinheads were applied with this method to a scale. The pinheads were in contact with the skin for ∼0.5 s before the experimenter raised the swivel arms to end the stimulus. Two identical lever apparatuses were used to apply the test stimuli, one for each forearm. Two experimenters were needed to conduct the experiment, each operating one lever apparatus. The order of the forearms receiving the test stimuli in each trial (either reference first or comparison first) was consistent for a given participant but counterbalanced across Frontiers in Human Neuroscience | www.frontiersin.org 3 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion FIGURE 2 | Force-controlled two-point stimulus apparatus. (A) Front view with a two-point applicator attached to magnets at the ends of the swivel arms. In this illustration, the applicator’s pinheads are separated by 42 mm. (B) Side view without applicator, illustrating the angle adjustment nut on one of the swivel arms. participants. Regardless of the order, in each trial, the stimuli were applied to the forearms sequentially. As one experimenter completed the first stimulus and raised the swivel arms away from the skin, the second experimenter initiated the stimulus to the other forearm. The inter-stimulus interval was ∼1 s. The two experimenters were trained to keep the application pace consistent between stimuli and across trials. The precise angles of the swivel arms were individually adjustable in order to match the slight change in thickness (and therefore height above the table) of the forearm along the proximal-distal axis. The experimenters adjusted the angles of the two swivel arms within each apparatus in order to ensure that the two pinheads contacted the skin simultaneously and with equal force, as reported by the participant. Qualification Criteria To ensure that participants’ baseline two-point distance perception was sufficiently accurate to perform the two-point distance comparison task, we compared participants’ baseline performance in the first testing block to two qualification criteria: the proportion of ‘‘comparison is longer’’ responses at the longest comparison distance (54 mm) should be ≥0.7, and at the shortest comparison distance (6 mm) should be ≤0.3. If a participant failed to meet either criterion, then we considered their baseline performance as unreliable. In this case, the participant did not proceed with the experiment, and their data were excluded from analysis. Psychometric Function Parameterization and Estimation of Point of Subjective Equality (PSE) For each of the three testing blocks for each participant, we fit to the data a sigmoidal cumulative normal function, which describes the proportion of trials at which the comparison distance, x, was reported as being longer than the reference distance: 9(x) = δ 2 + (1− δ) γ + (1− γ ) 1 σ √ 2π x∫ −∞ e−(t−µ) 2/2σ 2dt  This function has four free parameters: the mean (µ) and standard deviation (σ ) of the cumulative normal curve, a lapse rate (δ), and a y-intercept (γ ). We allowed γ to take on non-zero values, because the psychometric function for many participants did not fall completely to zero at the left tail. Using Bayesian parameter estimation, beginning with uniform prior probabilities over the four parameters, we calculated the joint (µ, σ, γ, and δ) posterior density.Wemarginalized this over δ and read out the mode of the (µ, σ, γ) posterior as the best-estimate of the participant’s psychometric function.We then extracted the comparison distance at which the psychometric function crossed 50% as the perceptual equivalent of the reference distance, i.e., the point of subjective equality (PSE). EXPERIMENT 2 In Experiment 2, we assessed the extent to which vibratory adaptation changed tactile sensitivity, by measuring participants’ 2IFC detection of force-calibrated Semmes-Weinstein monofilaments (a.k.a von Frey hairs; Timely Neuropathy Testing, LLC and Texas Medical Design, Inc., Dallas, TX, USA). We individually measured the application force produced by each filament with an analytical balance (model AB54-S/FACT, Mettler Toledo). Vibrotactile Adaptation Procedure The participant was seated in front of a table with the experimental apparatus concealed by an opaque black curtain. The participant’s right forearm rested comfortably in a supine position on a padded surface; the wrist was secured to a concave foam support. To mark the skin site for receiving vibratory adaptation, a circle of 19 mm diameter (the size of the adapting probe surface) was drawn with a fine-tipped pen on the volar forearm midway between the wrist and the elbow, and slightly Frontiers in Human Neuroscience | www.frontiersin.org 4 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion lateral to the proximal-distal midline; the center of the circle was at approximately the midpoint between the two reference points in Experiment 1. The adapting vibration was delivered via the plastic hemispherical surface of a JVP dome (Stoelting Co., Wood Dale, IL, USA; 19 mm diameter, 0.35 mm groove width). A mechanical arm holding the JVP dome was vibrated via the rotation of an attached eccentric motor (a NexxTech 1.98A DC motor whose axle we asymmetrically weighted, powered at 7.5V by DC power supply 1621A, BK Precision). A force sensor (Honeywell FSG15N1A) in contact with the end of the JVP dome shaft passed a voltage signal proportional to the contact force to an iMac computer via a USB board (NI USB-6210, 16-bit, National Instruments). A custom LabVIEW program monitored the force trace at 5000 samples/s. The program displayed the baseline indentation force and recorded the force waveform during vibration. To apply the adapting stimulus, the experimenter lowered the mechanical arm and pressed the JVP dome against the participant’s volar forearm at a perpendicular angle. Prior to and during the vibration, the experimenter adjusted the baseline indentation force to approximately 250 g. Post-experiment analysis on the force sensor data showed that the probe vibrated at 122 ± 5 Hz with a peak-to-peak force fluctuation of 125 ± 34 g (mean ± 1 SD; baseline force 245 ± 14 g). As soon as the adapting vibration ceased, the experimenter retracted the mechanical arm to remove the probe from the forearm. The experimenter then applied the monofilament test stimuli. The time between the offset of the adapting vibration and the application of the test stimuli was∼3 s. Experiment 2A To assess the strength of adaptation as a function of vibration duration, we measured participants’ ability to detect Semmes- Weinstein monofilament stimuli applied at the center of the adapted skin site in different adaptation conditions: (a) no-adaptation (NA); (b) 40 s initial adaptation without top-ups (A0); (c) 40 s initial adaptation plus a 3 s top-up vibration prior to each subsequent trial (A3); and (d) 40 s initial adaptation plus a 7 s top-up vibration prior to each subsequent trial (A7). The purpose of the top-ups was to prevent the adaptation effect from waning. After 20 practice trials with auditory feedback, participants completed the four testing blocks without feedback. Half of the participants completed the four blocks in the order NA- A0-A3-A7, and the other half in the order NA-A7-A3-A0. In the NA-A0-A3-A7 situation, participants took a 10 min break after completing NA, a 10 min break after completing A0, and a 15–20 min break after completing A3. In the NA-A7-A3- A0 situation, participants took a 10 min break after completing NA, a 15–20 min break after completing A7, and a 15–20 min break after completing A3. The breaks after A3 and A7 were longer than after NA or A0, because the A3 and A7 blocks lasted much longer due to the top-ups. The longer breaks were designed to allow participants to recuperate and their nervous systems to recover from possible long-lasting effects of adaptation. Each testing block had 100 2IFC trials. Each trial consisted of two intervals, separated by ∼1.25 s and demarcated by beeps. Simultaneously with one of the beeps, the skin was stimulated with a monofilament for∼0.5 s. By pressing one of two response keys with the left hand, the participant reported whether the stimulus occurred with the first or second beep. Monofilament force began at 0.07 g and was adaptively adjusted via a 2-down 1-up staircase procedure: If the participant answered correctly for two consecutive trials, the monofilament with the next-lower force was applied; if the participant answered incorrectly on any trial, the monofilament with the next-higher force was applied. This procedure converges towards the participant’s 71% correct detection threshold (Levitt, 1971). At the beginning of each adaptation block (A0, A3 and A7), the circled skin site received a 40 s vibration. Additionally, in the adaptation blocks with top-ups (A3 and A7), the circled site received a 40 s vibration when the participant returned from a break. Within each block, participants took a break after trials 33 and 66. For blocks NA and A0, which occurred relatively quickly, the break duration was 10 s. For blocks A3 and A7, which took much longer because of the top- ups, the break duration was 5 min to allow participants to recuperate. For each testing block, the participant’s 71% threshold was estimated by averaging the staircase reversal points in the last 50 of the 100 trials. In the rare circumstances in which the last 50 trials contained no reversal points and the participant consistently gave correct responses, so the staircase dropped to and continued at the lowest filament force, we used that force (0.008 g) as the estimated threshold. Experiment 2B To assess the spatial spread of vibrotactile adaptation, we used 40 s adaptation plus 3 s top-ups (protocol A3) and measured 2IFC monofilament detection at four distances from the center of adaptation. In addition to the circle drawn on the participant’s right volar forearm to indicate the site for vibrotactile adaptation, four dots were drawn at 0, 10, 15 and 20 mm from the center of the circle to mark the monofilament test sites. The dots were aligned along the proximal-distal axis of the forearm (Figure 1B). For half of the participants, the dots extended proximally, from the center of the circle towards the elbow; for the other half of the participants, the dots extended distally, from the center of the circle towards the wrist. Using interleaved 2-down 1-up staircases, we tested the four sites in consecutive trials in the order 0, 10, 15 and 20 mm from the center of the circle. For example, the 0 mm site was tested on trial 1, the 10 mm site on trial 2, the 15 mm site on trial 3, the 20 mm site on trial 4, and the 0 mm site again on trial 5. For all sites, the first trial used the 0.07 g monofilament. The force of the monofilament applied at each test site on subsequent trials followed the staircase procedure based on the participant’s responses at that site. For example, if the participant responded correctly on trials 1 and 5 on which the 0 mm site was tested, then the monofilament applied on Frontiers in Human Neuroscience | www.frontiersin.org 5 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion the next trial at that site (trial 9) went down to the next-lower force. After 20 practice trials with auditory feedback, each participant completed two testing blocks without feedback: a NA block and an adaptation (A3) block. Half of the participants completed the NA block first; the other half completed the A3 block first. Each block consisted of 200 trials (i.e., 50 trials at each of the four test sites). In the A3 block, prior to the first trial and every time the participant returned from a break, the circled skin site received a 40 s vibration. To prevent the adaptation effect from waning, the circled skin site received a 3 s top-up vibration prior to each of the subsequent trials. Participants took a 20 min break between testing blocks; within each block, they took a break after completing trials 33, 66, 100, 133 and 166 (break durations: NA block, 10 s after trials 33, 66, 133, 166, 5 min after trial 100; A3 block, 5 min after trials 33, 66, 133, 166, 10 min after trial 100). For each testing block, the participant’s 71% threshold at each test site was estimated by averaging the staircase reversal points in the last 25 of 50 trials at that site. In the rare circumstances in which the last 25 trials contained no reversal points and the participant consistently gave correct responses, so the staircase dropped to and continued at the lowest filament force, we used that force (0.008 g) as the estimated threshold. EXPERIMENT 3 In Experiment 3, we investigated the effects of vibratory adaptation on two-point distance perception. We applied the A3 vibrotactile adaptation protocol to the same 20 participants tested in Experiment 2B but on a different day. The participants compared two-point distances on the two forearms, as in Experiment 1, but with or without vibratory adaptation to the intervening skin between the reference points (Figure 1C). The test skin sites, exclusion criteria, and PSE estimation procedure were as described in Experiment 1. After practice, participants completed three testing blocks, a pre-adaptation (Pre) block without adapting vibration, an adaptation (A3) block, and a post-adaptation (Post) block without adapting vibration. The Pre and Post blocks were identical to the baseline testing blocks in Experiment 1. Participants took a 5 min break after the practice block and a 20 min break between testing blocks. During the Pre and Post blocks, participants took a 1-min break—and during the A3 block, a 5-min break—upon completing each quarter of the 90 trials (i.e., after completing trials 22, 45, 67). In the A3 block, prior to the first trial and every time the participant returned from a 5 min break, the skin midway between the two reference points (30 mm apart) on the right forearm received a 40 s adapting vibration. In addition, the same skin site received a 3 s vibration as a top-up adaptation prior to each subsequent trial, to prevent the adaptation effects fromwaning. The adapting probe was removed immediately from the skin when the adapting vibration ceased, and then the two pairs of test stimuli were applied to the forearms successively. The application order of the reference and comparison points was counterbalanced across participants: half of the participants received the reference points first in every trial, and the other half received the comparison points first in every trial. The time between the offset of the adapting vibration and the application of the reference points was ∼3 s for participants who received the reference points first, and ∼4 s for participants who received the comparison points first. Statistical Analyses We performed ANOVAs with type III sum of squares (and Greenhouse-Geisser correction to the degrees of freedom and the p-values in case of violation of sphericity) and two-tailed t-tests using SPSS Statistics version 20 (IBM) for Macintosh with an alpha level of 0.05. We performed two-tailed binomial proportion tests in R version 3.0.3. We used R version 3.0.3, companion to applied regression (car) package for post hoc one-way repeated-measures ANOVAs. For multiple post hoc pairwise comparisons, we used Bonferroni correction and reported p-values multiplied by the number of comparisons. RESULTS We undertook a series of three experiments to test for the presence of a tactile adaptation-induced repulsion illusion on the forearm. In a 2IFC task, participants compared the distances of two pairs of point-stimuli (reference vs. comparison) applied on their forearms successively, reporting which distance felt greater. The reference distance was fixed at 30 mm, and the comparison distance varied from 6 mm to 54 mm. The order of the reference and comparison distances was counterbalanced across participants. The PSE (i.e., the comparison distance reported as being greater than the reference distance 50% of the time) was extracted as a measure of participants’ perceived distance between the reference points. We measured baseline PSEs (Experiment 1) and PSEs following vibrotactile adaptation (Experiment 3). We used force-calibrated Semmes-Weinstein monofilaments to assess the efficacy of the adaptation protocol in reducing tactile sensitivity (Experiment 2). Baseline Distance Perception Was Unbiased and Stable In Experiment 1, we assessed the accuracy and stability of participants’ baseline two-point distance perception. Experiment 1 consisted of three identical testing blocks of the 2IFC distance- comparison test without adaptation. One participant reported that all comparison distances (6–54 mm) were greater than the reference distance (30 mm) in the third testing block; consequently, we could not reliably measure his psychometric curve or PSE for that block. We therefore excluded his data from all three blocks and analyzed the remaining 19 participants’ data. The average psychometric curves and estimated PSEs are shown in Figure 3 (Figure 3A: raw data. Figure 3B: psychometric function fits). The raw psychometric curves for some participants were noisy and crossed the y = 0.5 line multiple times, making it difficult Frontiers in Human Neuroscience | www.frontiersin.org 6 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion FIGURE 3 | Experiment 1 results. (A) Top: mean of raw data (N = 19 participants) for three NA blocks. Black, 1st block; brown, 2nd block; gray, 3rd block. Horizontal axis: comparison distance (mm). Vertical axis: proportion of trials in which the comparison distance (C) was perceived as greater than the reference distance (R = 30 mm). Dashed lines: Presp(C > R) = 0.5 and C = 30 mm. Error bars: ±1 SE (when error bars are not visible, it is because they are smaller than the data point circles). Bottom: for each testing block, the difference between 30 mm and the mean PSE, estimated by linear interpolation of the mean data (top). (B) Top: mean of the participants’ individual best-fitting psychometric functions. Error bars: ±1 SE. Bottom: difference between 30 mm and the mean of the PSEs extracted from the participants’ individual best-fitting psychometric functions. Error bars: ±1 SE. to extract individual PSEs directly from the raw data. Therefore, using the raw data we estimated only the across-participant mean PSE by linearly interpolating the mean response proportions (Figure 3A, top). The mean PSEs obtained in this fashion for the three baseline NA blocks were 29.38, 30.27 and 30.27 mm (Figure 3A, bottom). Binomial tests revealed that the proportion of trials in which participants judged the 30 mm comparison distance as longer than the 30 mm reference distance did not differ significantly from 0.5 for any block (p = 0.717, 0.828 and 0.828, for blocks 1, 2 and 3, respectively). Next, we used Bayesian curve fitting to estimate the psychometric functions and extract the PSEs of the individual participants. Each of the curves shown in Figure 3B (top) is an average of 19 individual best-fitting psychometric curves; the similarity of these three curves to those shown in Figure 3A (top) suggests that our curve fitting procedure provided a valid estimate of participant performance. The means (±1 SE) of the PSEs extracted from the participants’ individual best-fitting psychometric functions for the three blocks were 30.25 ± 1.08, 30.56 ± 1.13 and 29.39 ± 1.62 mm (Figure 3B, bottom). One-sample t-tests indicated that none of the PSEs differed significantly from the reference distance of 30 mm (block 1: t(18) = 0.227, p = 0.823; block 2: t(18) = 0.493, p = 0.628; block 3: t(18) = −0.373, p = 0.713), and a one-way repeated-measures Frontiers in Human Neuroscience | www.frontiersin.org 7 June 2017 | Volume 11 | Article 331 http://www.frontiersin.org/Human_Neuroscience http://www.frontiersin.org http://www.frontiersin.org/Human_Neuroscience/archive Li et al. Tactile Spatial Repulsion ANOVA indicated that the PSEs did not differ across blocks (F(1.485,26.727) = 0.458, p = 0.580). These results indicate that baseline two-point distance perception was unbiased and stable across testing blocks. Focal Vibration Caused a Reduction in Tactile Sensitivity Having found that participants’ baseline two-point distance comparison judgments were reliable, we next asked whether we could induce focal adaptation between the two reference points. In Experiment 2, we applied prolonged vibration locally to the skin on the reference arm, and we measured 2IFC monofilament detection thresholds as a function of vibration duration and distance from vibration center. In Experiment 2A, we found that vibration caused an elevation of monofilament detection thresholds (i.e., a reduction in tactile sensitivity) that increased with the duration of vibratio