Elsevier

Consciousness and Cognition

Volume 23, January 2014, Pages 1-11
Consciousness and Cognition

Can you tickle yourself if you swap bodies with someone else?

https://doi.org/10.1016/j.concog.2013.10.009Get rights and content

Highlights

  • Tickle ratings were lower for active relative to passive movements, even during BTI.

  • Self-tickle is robust across surprise and diminished predictability conditions.

  • The context-insensitivity of the attenuation supports active inference.

  • Evidence was found showing a version of the RHI exists within the BTI.

Abstract

The effect of the body transfer illusion on the perceived strength of self- and externally-generated “tickle” sensations was investigated. As expected, externally generated movement produced significantly higher ratings of tickliness than those associated with self-generated movements. Surprisingly, the body transfer illusion had no influence on the ratings of tickliness, suggesting that highly surprising, and therefore hard to predict, experiences of body image and first-person perspective do not abolish the attenuation of tickle sensations. In addition, evidence was found that a version of the rubber hand illusion exists within the body transfer illusion. We situate our findings within the larger debate over sensory attenuation: (1) there is an attenuation of prediction errors that depends upon the context in which sensory input is predicted (i.e., efference copy), and (2) sensory attenuation is a necessary consequence of self-generated movement irrespective of context (i.e., active inference). The results support the notion of active inference.

Introduction

Gibson’s (1962) work distinguishing touching from being touched provided insight into how perceptual awareness may be modulated by self-generated actions (see also Blakemore, Rees, & Frith, 1998). In Gibson’s experiments passive touch consisted of having a stimulus pressed upon the skin with no (intended) subsequent movement. The term “passive touch” has since been applied to situations where the stimulus is moved across the skin by an experimenter, or where the hand of the participant is guided in such a way that the input can resemble that of self-generated exploration, but with movement being externally generated (Symmons, Richardson, & Wuillemin, 2008).

Self-generated movements involve the intent to move. They also involve the planning of, the preparation for, and the execution of, movement (Grezes & Decety, 2001). Self-generated movements may also involve the production of a copy of the efferent signal (a copy of the motor commands), which can be used to predict the sensory consequences of motor commands (e.g., the flow of proprioceptive input one would receive upon executing a reach movement; see Blakemore et al., 1998, Blakemore et al., 2002; Blakemore, Wolpert, & Frith, 1998). It is thought that this so-called efference copy is used to make a comparison between the expected and actual consequences of movement. These predicted consequences are known as ‘corollary discharge’ and are thought to enable the fine-tuning and updating of action. In contrast, since externally generated movements do not involve motor commands, no efference copies are generated, and thus the consequences of those movements cannot be compared with a prediction.

The efference copy mechanism has been suggested as an explanation of how people distinguish externally generated movements from their self-generated movements (Blakemore et al., 2002). Specifically, if there is no (or minimal) discrepancy between the predicted and actual sensory input, attenuation of the sensory information occurs (Voss, Ingram, Haggard, & Wolpert, 2006); a nice example of this would be our inability to tickle ourselves. In contrast, since externally generated sensations are not predicted, they are not attenuated, and hence give rise to the ability of others to tickle us (Blakemore, Wolpert, & Frith, 2000).

A number of findings seem to favour the idea of such an efference copy mechanism. For example, when a slight delay or trajectory perturbation (i.e., varying the direction of a tactile stimulus’ movement as a function of the direction of a hand’s movement) is introduced to self-generated touch, the self-tickle effect is weakened (i.e., self-tickle becomes stronger), consistent with the notion of a fine-grained, spatiotemporally-sensitive prediction being at the heart of the tickle effect (Blakemore, Frith, & Wolpert, 1999). In brief, the motor control explanation for sensory attenuation equates percepts with prediction error rather than with predictions (since the intensity of the tickle reflects the magnitude of the prediction error), and explains sensory attenuation as the attenuation of prediction errors that depend upon the accuracy and precision with which sensory inputs are predicted (via the corollary discharge produced by the efference copy).

Attenuation of re-afferent signals has been investigated more generally, unearthing evidence suggesting that self-generated movement is associated with non-specific gating of a wider range of current input than just the specific input predicted on the basis of the efference copy. It is still an open question how wide the range of gated input is, and various phenomena have been studied, from the suppression of saccadic eye movements (Wurtz, 2008) to attenuation within the auditory system during self-generated ‘sound-making’ actions (Poulet & Hedwig, 2002). In somatosensory research, Gallace, Zeeden, Röder, and Spence (2010) argued that attenuating tactile stimuli during the execution of self-generated movements is advantageous as it prevents potentially irrelevant stimuli (e.g., a fly landing on one’s arm) from interfering with important movements (e.g., defending oneself from an attacker). Accordingly, Juravle and Spence (2011) have shown that the execution of self-generated movements during a juggling task suppresses task-irrelevant cutaneous inputs. They argue that comparison of the anticipated and actual states attenuates the unnecessary sensory information (cutaneous inputs in their experiments) and enhances the essential sensory information (kinesthetic information). Their finding is consistent with work (e.g., Hecht et al., 2008, Seki et al., 2003) showing that self-generated movements make it more difficult to notice tactile stimuli presented on a moving limb through non-specific gating of incoming information.

The idea of non-specific gating for the purposes of ensuring focus on task-relevant stimuli would explain the attenuation of self-generated tickling as a more general effect of self-generated movement, rather than as an efference copy mechanism exclusively associated with self-generated movement related to somato-sensation. Of relevance here is an earlier study (Claxton, 1975) which showed that perceived tickle is diminished when visual input of the dynamics of the tactile input is made available, suggesting that the tickle effect relies on general predictability of the stimuli, rather than on efference copies alone. Problematically, Claxton focused on passive touch conditions (similar to that of Gibson (1962), wherein an experimenter brushed a feather across a subject’s stationary arm) and did not test whether depriving participants of visual input had an influence on tickle ratings during self- and externally generated movements. This account thus explains findings of attenuation that are specific to the predictability of passively-received stimuli; it does not, however, help explain whether predictability plays a role in self- and externally generated movements and, as such, whether attenuation of sensory input occurs broadly (i.e., non-specifically) as a result of self-generated movements.

Recent work in theoretical neuroscience has challenged the idea of motor commands, and hence efference copies of motor commands (see Adams et al., 2012, Brown et al., 2011). This, simultaneously, puts pressure on the efference copy account of sensory attenuation and provides a unified framework for findings suggesting non-specific gating. This work is inspired by the free energy principle (Friston, 2010) which says that the brain is fundamentally engaged in minimising its free energy, or average prediction error, over time. According to this theory both perceptual inference and action simply minimise prediction error. In perception, prediction error is minimised by updating model parameters to fit the sensory input. For action, prediction error is minimised by using action (such as eye movement and palpation) to change the actual sensory input to fit with the predicted sensory input. Movement is not caused by descending motor commands but by predictions of the proprioceptive consequences of an intended movement. These predictions are fulfilled by classic reflex arcs, until the predicted proprioceptive state is attained. Crucially, this explanation for self-generated movement is thought to require the withdrawal of attention or precision from the sensory consequences of action. The argument here is: if ascending (proprioceptive or exteroceptive) prediction errors were not attenuated, they would subvert the predictions causing the intended movement. In other words, one has to suspend attention to the actual sensory consequences of movement to allow movement to unfold - otherwise, one would infer the intended movement was not occurring. If attention is conceived as gain modulation (encoding expected precision), the attenuation of sensory precision is entirely consistent with the withdrawal of attention from sensory inputs in the non-specific sense reviewed above. This furnishes another explanation for the tickle effect as a fundamental aspect of self-generated movement (Brown et al., 2013, Brown et al., 2011). In this “active inference” account, prior beliefs about an intended movement cause predictions and sensory attenuation. This is consistent with findings suggesting non-specific gating and makes the stronger prediction that it is not a limited range of sensory input but all current sensory input that is attenuated during action.

In summary, we can discern two alternative explanations for the tickle effect. First, there is an attenuation of prediction errors that depends upon specific predictions in the form of efference copy (Blakemore et al., 1999). Second, sensory attenuation is a necessary consequence of predictions about an intended movement. There is a crucial distinction between these accounts: In the first, sensory attenuation depends upon the context in which sensory input is predicted. For example, predicted somato-sensation depends on who is touched (the individual him or herself, or a stranger), how touch is delivered (incongruence between movement and tactile sensations delivered via delay or trajectory perturbation), and where touch is delivered (a real hand or a non-hand object). In this way, a highly surprising context for the touch should undermine predictions, and thus increase prediction error and intensify somato-sensation. In the second, active inference account, sensory attenuation is an inherent consequence of descending predictions and is context insensitive; here, a surprising context should have no influence on somato-sensation. In what follows, we seek to distinguish between these two accounts by changing the context in which predictions about sensory consequences were manipulated - namely through illusory manipulations of body image (i.e., the body transfer illusion and the rubber hand illusion). In this setting, if prediction errors determine percepts, we would expect the self-tickle effect to be sensitive to illusory context because the highly surprising experience of the illusions should undermine confident processing of the prediction and thereby increase the prediction error and intensify the tickle percept. Accordingly, if one swaps bodies with someone else, acquires new types of limbs, or experiences asynchronous touch, the ratings of tickliness should be higher. Alternatively, under active inference, sensory attenuation is a consequence of self-generated movement irrespective of context - and we would anticipate the self-tickle effect to be robust to illusory changes in body image and touch characteristics.

Section snippets

Method

The majority of rubber hand illusion-style experiments are designed to incorporate both synchronous and asynchronous touch conditions. Since the present study utilises the rubber hand illusion, it too was designed with these two conditions. In our complex experimental design, synchronous and asynchronous touch were administered in each of the two different body transfer illusion conditions. Undergoing this procedure (including intervening modification of the equipment) would take an inordinate

Group 1

The participants were 12 volunteers with a mean age of 30.5 years (SD = 8.0 years), six were female. They had no reported visual, tactile or muscular abnormalities. The data from two additional participants were deemed unreliable and were not included in the statistical analyses; the majority of their tickling scores were zero or very close to zero. Specifically, eight of 12 “tickle” ratings were zero or close to zero (i.e., <5 on a visual analogue scale from 0 [not tickly] to 100 [very tickly])

Stimulus and apparatus

The “tickle” device consisted of two triangular pieces of soft foam attached on either end of a wooden rod (length = 65 cm). The rod was mounted on a cradle so that it could be moved back-and-forth. The rod could be moved either by the participant using their right hand or, from the other end, by the experimenter. The foam made light contact with the participant’s, and experimenter’s, left palm.

A CCTV camera (Hikivision QCPass N816) was mounted on a bicycle helmet; participants sat directly

Design

Each group was subjected to a repeated-measures design. The independent variables were “movement type” which had two levels (self-generated vs. externally generated), “perspective” with two levels (participant’s vs. experimenter’s), and “limb type” which had three levels (real vs. baseball bat vs. none). “Group” acted as a between-subjects variable which had two levels (asynchrony in body transfer illusion vs. synchrony in body transfer illusion). The dependent variables were magnitude

Group 1

Participants were shown the experimental set-up and how the “tickle” device worked prior to testing. During the pre-experiment instructions, participants were also given an opportunity to feel the tickle of the piece of foam in both self- and externally generated conditions. They were told that they would feel the piece of foam move across the palm of their hand and that sometimes they would move it and sometimes the experimenter would move it. They were then told that their task was to rate

Dependent measures

After each condition, observers rated the tickliness of the stimuli on a visual analogue scale from 0 (not tickly) to 100 (very tickly). They also explicitly rated the strength of the body transfer illusion on a similar scale (i.e., 0 [no illusion] to 100 [very strong illusion]); it was explained to participants that “no illusion” meant that they felt as though they were occupying their own body, while “very strong illusion” should be marked if they felt as though they were occupying the

Tickle ratings

A 2 (perspective: participant vs. experimenter) × 2 (movement type: self-generated vs. externally generated) × 3 (limb type: real vs. baseball bat vs. none) × 2 (group: 1 vs. 2) mixed-model ANOVA on subjective ratings of how tickly the stimuli were revealed that the main effect of the movement type factor was significant [F(1, 21) = 8.19, p = .009], with a large effect size, i.e., partial eta-squared (ηp2 = .28). The tickles associated with externally generated movements (M = 45.35, SE = 4.51) were consistently

Discussion

The tickle effect refers to our inability to tickle ourselves. The main issue explored in this study was whether the tickle effect is influenced by changes in context, here in body image and spatiotemporal touch characteristics. Remarkably, the tickle effect is not abolished by the dramatic changes in context induced by the body transfer illusion and rubber hand illusion. Even as participants shift their first-person perspective to someone else’s, or experience having a baseball bat as a hand,

Conclusion

We asked “can you tickle yourself if you swap bodies with someone else?” The short answer is “no”. We show that the attenuation of self-generated tickle sensations is remarkably robust across contexts that introduce various kinds of surprise and diminished predictability. This is consistent with findings of non-specific sensory gating during self-generated movement and supports the active inference hypothesis. This is remarkable because it suggests that body image and first-person perspective

Acknowledgments

This research was supported by a Monash University Small Grant Research Support Scheme grant (G.V.), a Monash University Emerging Research Excellence Fellowship (G.V.), an Australian Research Council Future Fellowship (J.H.), and a Monash University Research Accelerator grant (M.S.). We wish to thank the editor and anonymous reviewers for numerous helpful suggestions.

References (33)

  • S.-J. Blakemore et al.

    Central cancellation of self-produced tickle sensation

    Nature Neuroscience

    (1998)
  • S.-J. Blakemore et al.

    Why can’t you tickle yourself?

    NeuroReport

    (2000)
  • M. Botvinick et al.

    Rubber hands ‘feel’ touch that eyes see

    Nature

    (1998)
  • H. Brown et al.

    Active inference, sensory attenuation, and illusions

    Cognitive Processing

    (2013)
  • H. Brown et al.

    Active inference, attention and motor preparation

    Frontiers in Psychology

    (2011)
  • G. Claxton

    Why can’t we tickle ourselves?

    Perceptual and Motor Skills

    (1975)
  • Cited by (14)

    View all citing articles on Scopus
    View full text