Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling

Article

Behavioral and Cortical Correlates of SelfSuppression, Anticipation, and Ambivalence in Rat Tickling Highlights d

Self-touch suppresses vocalizations and cortical excitation

d

Self-touch suppression is rescued by blocking cortical inhibition

Authors Shimpei Ishiyama, Lena V. Kaufmann, Michael Brecht

Correspondence

d

Rats show ambivalent response to tickling

[email protected]

d

Layer 5 somatosensory cortex represents tickle anticipation

In Brief Ishiyama et al. perform in vivo electrophysiology in rat somatosensory cortex. Although allo-touch evokes vocalizations and cortical excitation, selftouch suppresses them. Rats show ambivalent response to tickling. Layer 5 somatosensory neurons represent tickle anticipation.

Ishiyama et al., 2019, Current Biology 29, 1–12 October 7, 2019 ª 2019 The Authors. Published by Elsevier Ltd. https://doi.org/10.1016/j.cub.2019.07.085

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Current Biology

Article Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling Shimpei Ishiyama,1,2 Lena V. Kaufmann,1 and Michael Brecht1,3,4,*

1Bernstein Center for Computational Neuroscience Berlin, Institut fu €t zu Berlin, Philippstraße 13, Haus 6, ¨ r Biologie, Humboldt-Universita 10115 Berlin, Germany 2Institut fu €tsmedizin der Johannes Gutenberg-Universita €t Mainz, Duesbergweg 6, 55128 Mainz, Germany ¨ r Pathophysiologie, Universita 3NeuroCure Cluster of Excellence, Humboldt-Universita €t zu Berlin, Charite platz 1, 10117 Berlin, Germany 4Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.cub.2019.07.085

SUMMARY

The relationship between tickling, sensation, and laughter is complex. Tickling or its mere anticipation makes us laugh, but not when we self-tickle. We previously showed rat somatosensory cortex drives tickling-evoked vocalizations and now investigated self-tickle suppression and tickle anticipation. We recorded somatosensory cortex activity while tickling and touching rats and while rats touched themselves. Allo-touch and tickling evoked somatotopic cortical excitation and vocalizations. Self-touch induced wide-ranging inhibition and vocalization suppression. Self-touch also suppressed vocalizations and cortical responses evoked by allo-touch or cortical microstimulation. We suggest a global-inhibition model of self-tickle suppression, which operates without the classically assumed self versus other distinction. Consistent with this inhibition hypothesis, blocking cortical inhibition with gabazine abolished self-tickle suppression. We studied anticipation in a nose-poke-for-tickling paradigm. Although rats nose poked for tickling, they also showed escaping, freezing, and alarm calls. Such ambivalence (‘‘Nervenkitzel’’) resembles tickle behaviors in children. We conclude that self-touchinduced GABAergic cortical inhibition prevents self-tickle, whereas anticipatory layer 5 activity drives anticipatory laughter. INTRODUCTION Tickling is an idiosyncratic form of social touch. What makes tickling so distinct from other forms of touch is the fact that it evokes laughter. This phenomenon is also referred to as ‘‘gargalesis’’ [1] and distinguished from knismesis, i.e., slightly itchingtickling touch. For thousands of years, humans have been fascinated by ticklishness; Aristotle posed the question of why we cannot tickle ourselves and whether we are more ticklish when

we anticipate it [2]. To answer such questions, we need to understand the evolutionary neurobiology of ticklish touch and how expectations shape self-touch and ticklish sensations. A breakthrough finding in tracing the evolutionary history of tickling came from Panksepp and Burgdorf’s work on rat ticklishness [3, 4]. The authors showed that rats emit 50-kHz vocalizations in response to tickling by human, that tickling is rewarding to rats, and that young rats are much more ticklish than adults— by all standards a remarkable set of parallels to human ticklishness. Panksepp and Burgdorf also pointed out behavioral links between ticklish touch and play [5]. Specifically, they showed that blocking sensations of the dorsal surface, a ticklish skin part in rats, strongly interferes with play fighting and play initiation [6]. In a subsequent study, Burgdorf et al. showed that the rewarding characteristics of tickling are dependent on the integrity of the dopaminergic system [7]. Inspired by Panksepp and Burgdorf’s work on rat tickling and by their finding that play behavior is triggered by somatosensory stimulation in the dorsal skin, we investigated the role of trunk somatosensory cortex in rat ticklishness [8]. We found tickling evokes strong responses in rat trunk somatosensory cortex. Remarkably, such responses are suppressed by anxiogenic conditions, in accordance with the finding that, in humans [9] and rats [3], ticklishness is suppressed by fear or anxiety. Deep-layer somatosensory cortex neurons discharge prior to and during 50-kHz vocalizations, and electric stimulation of deep, but not superficial, somatosensory cortex layers evokes vocalizations. Thus, these findings suggest a causal role of the deep layers of somatosensory cortex in rat ticklish laughter. In line with the behavioral evidence, we found a correlation of somatosensory cortex response patterns to tickling and to play behaviors, suggesting a neural link between tickling and play. The observation that we cannot tickle ourselves is one of the oldest issues in the human thought about tickling. Both Aristoteles [2] and Darwin [9] suggested that the self-tickle suppression might be related to the predictability of self-touch. A number of more recent studies have investigated the mechanisms that prevent self-tickling. Behavioral work using a tickling machine [10] for indirect self-tickling indicated that, indeed, self-tickling inhibition was reduced if self-touch was applied more indirectly. Subjects felt ticklish when they were told that they were tickled

Current Biology 29, 1–12, October 7, 2019 ª 2019 The Authors. Published by Elsevier Ltd. 1 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

by a tickling robot, indicated that tickling does not require the concept of a tickling agent [11]. Furthermore, functional imaging suggested relatively detailed mechanisms preventing self-tickle in humans [12, 13]. These authors found a reduced activation of somatosensory cortex due to self-tickling compared with allotickling. Specific cerebellar activation patterns in the two tickling conditions suggested that the cerebellum is involved in the cancellation of self-tickling responses. In our current study, we ask three questions about rat ticklishness. First, we ask how the somatosensory cortical representation of self-touch differs from the representation of allotouch and tickling. In particular, we are interested whether physiological differences between self-touch, allo-touch, and tickling could help explain our inability to evoke laughter by self-tickling. Second, we try to determine the rats’ attitude toward tickling in a self-initiated tickling paradigm. To this end, we give rats an opportunity to self-initiate tickling interactions by nose poking. We then analyze behavioral responses concomitant with such tickling initiation and tickling expectation. Third, we study the role of anticipation in rat tickling. In line with self-tickle suppression, self-initiated tickling could lead to reduced ticklishness if suddenness and unpredictability increase ticklish sensation, as claimed by Aristotle [2] and Darwin [9]. Thus, we varied nose-poke-to-tickling latency to investigate whether temporal uncertainty affects response to subsequent tickling stimuli. We also used this paradigm (in which the rat knows that it will be tickled) to assess the presence of tickle-anticipatory cortical activity. We found that self-touch induces inhibition in trunk somatosensory cortex, which might underlie the reduced rate of vocalizations during self-touch. The self-touch suppression of allo-touch-evoked vocalization was abolished by disinhibition of trunk somatosensory cortex. Our self-initiated tickling paradigm confirmed that tickling is rewarding for rats, but concomitant escape, freezing, and fear call responses also revealed a great deal of ambivalence about ticklishness in rats. The selfinitiated tickling paradigm also suggested that layer 5a neurons might be drivers of anticipatory laughter. RESULTS Our study builds on our earlier finding that deep-layer neurons in trunk somatosensory cortex drive tickling-evoked vocalizations. These results underlie our new experiments, in which we compared behavioral and neural responses in rat trunk somatosensory cortex to self-touch, allo-touch, and tickling. In our investigation, we used grooming as a model for self-touch, which offers four major advantages: (1) rats touch themselves during grooming, (2) rats perform such self-touch movements spontaneously without further training, (3) grooming movements occur at a high rate, and (4) grooming behaviors systematically cover the body surface [14]. A potential disadvantage of the grooming model of self-touch is that such behaviors are stereotyped and serve a specific self-cleaning purpose, i.e., they are not under experimental operant control and might not mirror every aspect of human self-touch. We then compared behavioral and physiological responses to grooming, to allo-touch, and to tickling applied by a human experimenter, as described previously [8]. 2 Current Biology 29, 1–12, October 7, 2019

Comparison of Self-Touch, Allo-Touch, and Tickling Responses We identified episodes of self-touch, allo-touch, or tickling (Figure 1A). Next, we quantified vocalization rate during self-touch (right trunk grooming), allo-touch, and tickling on the dorsal trunk (Figure 1B). Interestingly, self-touch, allo-touch, and tickling are associated with drastically different patterns of vocalizations. In all cases, the ongoing rate of vocalization was slightly lower than 1 Hz. Although self-touch was associated with a decrease in vocalization rate, allo-touch and even more so tickling were associated with a sharp increase in vocalization rate (Figure 1B). Next, we quantified cortical responses of the left-trunk somatosensory cortex to contralateral self-touch, allo-touch, and tickling. Population histograms of trunk somatosensory single-unit response are represented in Figure 1C. Similar to the vocalizations, the neuronal responses varied between the three forms of touch: contralateral self-touch resulted in a suppression of neuronal activity, whereas allo-touch and tickling evoked strong excitatory responses. More specifically, self-touch responses differed in two ways from allotouch or tickling responses: (1) there was no brisk onset response after contact and (2) a slow, long-lasting reduction in response rate after stimulus onset was seen. Both of the differences are suggestive of the action of cortical inhibition in the trunk somatosensory cortex. We also analyzed cell by cell how cortical response to self-touch, allo-touch, and tickling were related. We computed response indices ranging from +1 for completely excitatory responses to 0 for no response to 1 for completely inhibitory responses for each cell during self-touch, allo-touch, and tickling. We found that inhibitory responses predominated for self-touch and excitatory responses predominated for allo-touch and tickling, but cellular responses were not systematically related (response indices: allo-touch versus contralateral trunk grooming, 0.153 and 0.004 to 0.327 versus 0.038 and 0.443 to 0.143, median and interquartile ranges; p < 0.001, signed-rank test; Figure 1D). When we analyzed the relation of cellular responses in trunk cortex to tickling and self-touch, we reached the same conclusion. Inhibitory responses predominated for self-touch and excitatory responses predominated for tickling, but cellular responses were only weakly correlated (Figure 1E). In contrast, responses to allo-touch and tickling were closely related (Figure 1F), in line with our earlier work [8]. Spike response index for tickling was significantly larger than response index for allo-touch (p < 0.001; signed-rank test; Figure 1F), consistent with previous study [8]. The data presented in Figure 1 suggest three conclusions: (1) a systematic relationship between vocalizations and neuronal responses in trunk cortex—vocalizations and cortical responses were suppressed during self-touch, increased during allo-touch, and strongly increased during tickling; (2) a distinction of cellular responses to self-touch on one hand and allo-touch and tickling on the other hand; and (3) a similarity of cellular responses to allo-touch and tickling. Trunk Cortical Inhibition Evoked by Self-Touch on Different Body Parts These initial results prompted us to further characterize behavioral and neuronal responses to self-touch. We quantified vocalization rates associated with head grooming and omnilateral scratching (Figure 2A). In line with the results shown in Figure 1, we observed a suppression of vocalizations during both

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Figure 1. Vocalizations and Trunk Somatosensory Cortex Responses Associated with Self-Touch, Allo-Touch, and Tickling (A) Schematics of self-touch (trunk grooming), allotouch (dorsal trunk), and tickling (dorsal trunk). (B) Population ultrasonic vocalization (USV) peristimulus time histogram (PSTH) during self-touch (right trunk grooming; 18 phases; 2 rats), allo-touch (44 phases; 3 rats), and tickling on the dorsal trunk (30 phases; 3 rats). Data are binned into 0.5-s width. Width of light gray boxes indicates median duration of phases. Phases with previous phases that occurred between 10 and 0 s were excluded to clarify the time course of the effect. (C) Population PSTHs of firing rate in all trunk somatosensory cortex neurons recorded during grooming on the contralateral trunk (37 cells; 15 phases; 6 rats), allo-touch (61 cells; 39 phases; 6 rats), and dorsal tickling (75 cells; 37 phases; 6 rats). Phases with previous phases that occurred between 10 and 0 s were excluded to clarify the time course of the effect. (D) Top: relationship of cellular responses to dorsal allo-touch versus responses to contralateral trunk grooming quantified by response indices (0.153 and 0.004–0.327 versus 0.038 and 0.443– 0.143; median and interquartile ranges; +1, complete excitation; 0, no change; 1 complete inhibition; see STAR Methods; p < 0.001, signedrank test). Orange cross indicates the mean value. Histograms indicate number of cells. Bottom: statistical parameters are shown: ncell, number of cells; r, Pearson’s correlation coefficient; p, p value of the correlation coefficient. (E) Same as (D) but for responses to dorsal tickling versus responses to contralateral trunk grooming (0.15 ± 0.03 versus 0.13 ± 0.04; p < 0.001, paired t test). Data are fitted with a line. (F) Same as (D) but for responses to dorsal tickling versus responses to dorsal allo-touch (0.195 and 0.002–0.388 versus 0.153 and 0.004–0.327; median and interquartile range; p < 0.001, signedrank test). Data are fitted with a line.

behaviors (Figure 2B). Next, we analyzed neuronal responses in the left trunk cortex to the same behaviors. Again, we observed self-touch responses differed from allo-touch or tickling responses, in that there were (1) no brisk onset responses to touch and (2) a slow, long-lasting reduction in firing rate after event onset. These observations are suggestive of the action of cortical inhibition. In population peristimulus time histograms, we found that trunk cortical neurons were inhibited by head grooming and showed no response or slight inhibition to scratching (Figure 2C). Most interestingly, we found that the response indices of self-touch responses to head and contralateral trunk grooming (Figure 2D) and to contralateral trunk grooming and scratching (Figure 2E) were significantly correlated. Analysis on spike response indices and cortical layers demonstrated that tickling

and allo-touch activated all layers but layer 6, and self-touch resulted in inhibition in all layers but layer 6 (data not shown). In summary, we observed three findings on relationship in cortical response to touch: (1) allo-touch and tickling responses are correlated in strength (Figure 1F); (2) self-touch responses are correlated in strength (Figures 2D and 2E); and (3) self-touch and allo-touch-and-tickling responses were correlated not in strength (Figures 1D and 1E). This pattern of results suggests that self-touch (head grooming, trunk grooming, and scratching) responses on one hand and allo-touch/tickling responses on the other hand might be driven by two distinct sources. Self-Touch Suppression of Allo-Touch-Evoked Vocalizations, Microstimulation-Evoked Vocalizations, and Allo-Touch-Evoked Cortical Responses We wondered whether these distinct properties of self-touchinduced inhibition might account for the inability to self-tickle Current Biology 29, 1–12, October 7, 2019 3

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Figure 2. Widespread Inhibition in Trunk Somatosensory Cortex upon Self-Touch (A) Schematics of head grooming and scratching. (B) USV PSTH for head grooming (19 phases; 3 rats) and omnilateral scratching (63 phases; 5 rats). Data are binned into 0.5-s width. Width of light gray boxes indicates median duration of phases. Phases with previous phases that occurred between 10 and 0 s were excluded to clarify the time course of the effect. (C) Population PSTH of firing rate in all trunk somatosensory cortex neurons recorded during grooming of the head (41 cells; 14 phases; 6 rats) and during scratching (55 cells; 59 phases; 6 rats). Phases with previous phases that occurred between 10 and 0 s were excluded to clarify the time course of the effect. (D) Top: relationship of spike response index for head grooming versus contralateral trunk grooming ( 0.07 and 0.294–0.081 versus 0.038 and 0.443–0.143; median and interquartile range; p = 0.188, signed-rank test). Data are fitted with lines. Orange cross indicates the mean values. Histograms indicate number of cells. Bottom: statistical parameters are shown. (E) Same as (D) but for scratching versus contralateral trunk grooming ( 0.046 and 0.385–0.115 versus 0.038 and 0.443–0.143; median and interquartile range; p = 0.834, signed-rank test).

4 Current Biology 29, 1–12, October 7, 2019

and thus addressed this question in behavioral, cortical stimulation, and physiological experiments (Figure 3). Specifically, we assessed the effects of self-touch on touch-evoked vocalizations. To this end, we compared allo-touch-evoked vocalizations out of self-touch and peri-self-touch (Figure 3A). To minimize interruption of self-touch by allo-touch, rats were habituated to tickling so that light one-finger tapping on the lateral trunk was sufficient to evoke ticklish vocalizations. During the experiment, the experimenter’s hand was kept in the tickle box or slowly approached the rats without being noticed, and the rats were tapped on the trunk as soon as they initiated self-touching. Population analysis demonstrated that, surprisingly, tapping-evoked vocalizations were significantly reduced during self-touch (3.97 ± 0.24 versus 2.13 ± 0.37 Hz, mean ± SEM; p < 0.001, paired t test; Video S1; Figure 3B). Note that self-touch did not completely suppress all allotouch-evoked vocalizations but that it merely led to a reduction in the rate of allo-touch-evoked calls. This finding suggests that selftickle suppression may not be based on the distinction between self- and allo-touch, as it is often assumed. If the brain were to compute this self- versus allo-touch distinction, allo-touch should remain effective in evoking vocalizations during self-touch. Instead, this behavioral result is consistent with the physiologyderived hypothesis that self-touch-induced global inhibition suppresses vocalizations. Indeed, not only trunk grooming but also head grooming significantly suppressed vocalizations evoked by trunk tapping (peri-trunk-groom versus control, 2.7 ± 0.4 versus 4.4 ± 0.3 Hz, p < 0.001, 12 recordings from 4 rats; peri-headgroom versus control, 1.95 ± 0.39 versus 4.0 ± 0.2 Hz, p < 0.001, 9 recordings from 2 rats; mean ± SEM, paired t test; Figure S1). To investigate the underlying physiological mechanisms of self-touch-induced suppression of vocalizations, we turned to microstimulation experiments in the somatosensory cortex (Figures 3C and 3D). We microstimulated the trunk cortex out of (Figure 3C, left) and during self-touch (i.e., grooming, scratching, and paw chewing; Figure 3C, right). We found electrically evoked vocalizations were significantly reduced when applied during selftouch (1.66 ± 0.28 versus 0.85 ± 0.17 Hz, mean ± SEM, paired t test; Figure 3D). These data indicate that self-touch-induced suppression of vocalization occurs in or downstream of trunk somatosensory cortex. The results argue against the idea that selftouch-induced suppression of vocalization acts upstream of trunk somatosensory cortex, in which case vocalizations evoked by somatosensory cortex stimulation should not have been affected. To ascertain that the reduced number of vocalizations is indeed a result of cortical inhibition, we recorded trunk somatosensory cortex neurons when applying allo-touch out of trunk-groom and peri-trunk-groom (Figures 3E and 3F). We found that cortical neurons responded more strongly when tapping alone was applied (Figure 3E, left) than when we tapped the animal peri-grooming (Figure 3E, right). This reduction in tactile responsiveness during self-touch was significant across the population of units we recorded (Figure 3F). We conclude that self-touch reduces allotouch-evoked vocalization, vocalizations evoked by cortical microstimulation, and allo-touch-evoked cortical responses. Trunk Cortical Inhibition Is Necessary for Self-Tickle Suppression Our data so far provided a tight link between self-touch-induced cortical inhibition and self-tickle suppression. Still, we wondered

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Figure 3. Self-Touch Suppresses TappingEvoked Vocalizations, MicrostimulationEvoked Vocalizations, and Tapping-Evoked Cortical Responses (A) Representative USV PSTHs for tapping on the omnilateral trunk out of self-touch (left) and periself-touch (right), averaged over 28 and 12 phases, respectively. Light gray boxes represent median durations of tapping series. Data are binned into 50-ms width. (B) Population tapping-evoked USV rates out of self-touch (control) and peri-self-touch. Data are mean ± SEM 9 recordings from 2 rats; p < 0.001 (paired t test). (C) Representative USV PSTHs for microstimulation (100 Hz; 189 ± 4 mA; 2 s) in layer 5b trunk somatosensory cortex out of self-touch (left) and during self-touch (right), averaged over 36 and 9 phases, respectively. Data are binned into 50-ms width. (D) Population microstimulation-evoked USV rates out of self-touch (control) and during self-touch. Data are mean ± SEM 12 stimulation sites (2 L4; 3 L5a; 4 L5b; 3 L6; each layer from 2 animals; stimulation amplitudes: 154 ± 26 mA); p = 0.010 (paired t test). Stimulation amplitudes were consistent within each pair. USV rates were calculated from entire time where stimulation was delivered. (E) Representative spike PSTHs of a L5a trunk somatosensory neuron for tapping out of selftouch (left) and peri-grooming (right), averaged over 27 and 18 phases, respectively. Data are binned into 100-ms width. (F) Relationship of spike response index for control tapping versus peri-groom tapping (0.348 ± 0.025 versus 0.134 ± 0.031; mean ± SEM; p < 0.001, paired t test; magenta: 30 single units, green: 123 multi-units). Orange cross indicates the mean values. Histograms indicate the number of units. See also Figure S1 and Video S1.

whether trunk cortex inhibition is indeed necessary for self-tickle suppression and what is the precise cellular nature of these inhibitory signals. To answer this question, we locally blocked GABAergic transmissions in trunk cortex (Figure 4A, left) by injecting 250–500 nL of the GABAA receptor antagonist Gabazine (SR-95531), resulting in 12.5–100 pmol injection (Figure 4A, right). Gabazine acts as an allosteric inhibitor of the opening of GABAA channels. As before, we then applied allo-touch out of trunk-grooming and peri-trunk-grooming. There appeared to be some behavioral effect of the gabazine injection on the animals’ grooming behavior, i.e., we observed that the animals groomed significantly more frequently (1.8 times; p = 0.037, paired t test) the body side represented by the disinhibited cortex. When we then applied the touch paradigm to the body side ipsilateral to the injection site (which is represented by intact, i.e., uninjected trunk cortex), we observed a self-touchinduced suppression of vocalizations, as expected (Figure 4B). When we applied our touch paradigm to the body side

contralateral to the injection site (which is represented by disinhibited trunk cortex), we observed that self-touch-induced suppression of vocalizations was abolished, however (Figure 4C). This difference between self-touch-induced suppression of vocalization between body sides with intact and disinhibited trunk cortex was significant across experiments, and the removal of self-touch-induced suppression of vocalization was complete (Figure 4D). Importantly, we did not observe such a blockade of self-touch-induced suppression of vocalization in control experiments, in which we injected saline into trunk cortex (Figure S2). We conclude that GABAergic transmission in the trunk somatosensory cortex mediates self-touch-induced suppression of vocalizations. Behavioral Ambivalence in Self-Initiated Tickling Interactions Conditioned place-preference techniques [15], self-initiated tickle interaction [16], and the occurrence of freudenspru¨nge (joy jumps) after tickling [8] have suggested that rats enjoy being tickled. In order to better understand the rat’s attitude toward Current Biology 29, 1–12, October 7, 2019 5

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

tickling and to determine how expectation modulates rat ticklishness, we employed a nose poke paradigm. Animals were trained to nose poke to be tickled, which allowed the animal to selfinitiate tickling interactions. Animals did not spontaneously nose poke prior to training. We placed the animals in a box with a nose poke hole and a beeper (emitting a 0.5-s tone) triggered by a beam breaker in the nose poke hole (Figure 5A). After the beep, the animal was tickled with varying delays. All rats acquired the nose-poking behavior and increased the number of nose pokes over training sessions (Figure 5B), confirming earlier conclusions that tickling is rewarding. Surprisingly, however, the animals also showed other behaviors that we did not expect. The animals often adopted a crouched body position and entered brief episodes of immobility, i.e., freezing behavior (duration of freezing = 1.23 ± 0.10 s, n = 154 nose pokes; freezing probability after nose poke = 69%; Figure 5C; Video S2). After the nose poke but prior to freezing, the rats did show brief spurts, which could be interpreted as escape behaviors. Both escape and freezing behaviors were evident as high and close-to-zero speed deflections, respectively, when animal’s speed was quantified across the trials (Figure 5D). The occurrence of freezing, a post-threatdetection behavior [17], is particularly odd, and we therefore analyzed how such freezing episodes related to the rest of the trial and the animal’s vocalization. Figure 5E shows the analysis of a session with numerous freezing episodes aligned to freezing onset and sorted by the length of the freezing episode. Typically, a tickle interaction was initiated by a nose poke, often followed first by a brief escape spurt and then by an episode of freezing lasting until the onset of tickling. The animal did not reinitiate freezing after tickling but would occasionally freeze after failed nose pokes (not deep enough; freezing probability after failed nose-poke = 7%). Note that such failed nose pokes did not induce the beeping sound or tickling. The rats often exhibited failures, even after the acquisition of the operant behavior (nose poke success rate = 44%). Occasionally, the animals stayed still in front of the hole without poking it, as if they had hesitated to initiate tickling interaction, which supports the suggested ambivalence in self-initiated tickling. The animal’s vocalization rate varied systematically (Figures 5E and 5G) across the trial. It was moderate prior to the nose poke, very low during nose poke and freezing, and then ramped up to high levels during tickling. In line with the idea that rats perceived the situation as threatening, we also occasionally observed 22-kHz alarm calls after the nose poke (Figure 5F; Video S2). Population data confirmed these systematic variations in the vocalizations (during nose poke beep: 0.11 Hz and 0.00–0.37 Hz, median and interquartile ranges; freezing: 0.17 ± 0.05 Hz, mean ± SEM; tickling: 4.34 ± 0.20, mean ± SEM, n = 9 recordings; Figure 5G). We conclude that rats learn to self-initiate tickling interactions by nose poking. At the same time, they showed escape spurts, Figure 4. Cortical Disinhibition Disrupts Self-Touch Suppression of Ticklishness (A) Left: cytochrome oxidase stain of a coronal trunk somatosensory cortex section (D, dorsal; M, medial; red fluorescence, DiI trace of injection pipette; arrowhead, pipette tip). Right: schematic of gabazine (SR-95531, GABAA receptor antagonist) injection is shown. (B) Population USV peristimulus histograms for tapping (left) and perigroom tapping (right) on the control side of the trunk. Light gray boxes indicate median duration of tapping phases (averaged over 94 and 41

6 Current Biology 29, 1–12, October 7, 2019

phases, respectively; 6 recordings from 2 rats). Data are binned into 500-ms width. (C) Same as (B) but for the disinhibited side of the trunk (purple), averaged over 101 and 59 phases, respectively. (D) Population USV rate for peri-groom tapping normalized to USV rate for tapping on the control and disinhibited side of the trunk (0.61 ± 0.05 versus 0.95 ± 0.08, mean ± SEM; p = 0.013, paired t test). See also Figure S2.

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

alarm calls, and freezing behavior, characterized by crouched posture, immobility, and a low vocalization rate after nose poking.

Figure 5. Ambivalence in Self-Initiated Tickle Interactions: Escape, Freezing, and Fear Calls (A) Schematic of nose poke box. The front wall is cut for visualization. (B) Learning curve for nose-poke-initiated tickling paradigm: numbers of successful nose poke over number of operant conditioning training for 30 min for 4 rats. Data after training period, i.e., experiments with implantation, are not shown. (C) Photos of nose poking (left) and freezing (right); note the crouched body posture during freezing. (D) Top: color coding of behavioral events during a self-initiated tickling episode. Bottom left: speed trajectory of an escape (peak in speed) and freezing response (green box) after nose poke is shown (red line). Bottom right: superimposed speed trajectories of a set of trial from session with freezing responses are shown (light gray, 46 phases). Width of the green box represents mean duration of freezing. Dark gray line indicates the mean speed. Red dashed lines display minimal and maximal duration of freezing.

Anticipatory Neural Responses to Self-Initiated Tickling Interactions How does neuronal activity in trunk somatosensory cortex evolve in self-initiated tickling interactions? We observed marked laminar differences in the temporal evolution of responses during self-initiated tickling. As in a raw trace recorded in layer 4 trunk somatosensory cortex, cellular discharges were locked to the actual tickling (Figure 6A). A different pattern was observed in raw traces recorded in layer 5a trunk somatosensory cortex: here, cellular discharges ramped up during freezing and reached the maximum rate prior to tickling but are also sustained during tickling (Figure 6B). An across-trial analysis confirmed such response differences between the layer 4 and the layer 5a cell (Figures 6C and 6D). Interestingly, this ramped-up response was seen only after successful nose pokes. After nose poke failures, which did not induce a beep and tickling, the animals did occasionally show freezing, but no ramp up of the layer 5a cell activity was seen (Figure 6D). Indeed, the population analysis showed that the ramp up of layer 5a activity is only seen after successful (Figure S3A, right), but not for aborted, nose pokes (Figure S3B, right). An acrosscell population analysis strengthened the idea that layer 4 activity follows tickling (Figure 6E), whereas layer 5a ramps up of firing up to 2 s prior to tickling onset (Figure 6F). On the other hand, layer 5b responses increased at the onset of tickling and did not show an anticipatory response like layer 5a neurons (28 cells; data not shown). Thus, we observe putatively sensory signals in layer 4, the cortical input layer, and putatively anticipatory signals in layer 5a, the cortical output layer. These data are noteworthy (1) because of the intracortical change of signal characteristics, (2) because of the major role of anticipation in ticklish laughter, and (3) because we showed previously that layer 5 stimulation can drive tickling-related vocalizations [8]. We also studied anticipation effects in another paradigm, in which we approached the rat with the tickling hand very slowly over a 10-s period (Figure S4). We found that this slow approach indeed caused a ramping up of the animal’s vocalizations prior to tickling (Figure S4B), in contrast to low vocalization rate during post-nose-poke freezing. At the same time, we observed a ramping up of cortical responses, which had a very similar time course (Figure S4C). Different from the nose poke paradigm, ramping up of cortical activity in this paradigm seemed to occur in most cortical layers (Figure S4D). (E) Representative raster plot of vocalizations (individual calls denoted as gray tick marks) and behavioral events (color coded; see legend in D) around the freezing responses. To allow a better visualization, we aligned trial to freezing onset and sorted them according to freezing duration. (F) Spectrogram (gray) and behavioral events (top, color coded) for a postnose-poke fear call; the nose poke beep sound (left) and 50-kHz vocalizations during tickling (right) are also visible in the spectrogram. (G) Population vocalization rates for phases in nose poke experiments. Boxplots are medians and interquartile ranges. Break versus freezing, p = 0.026; break versus tickling, p < 0.001; signed-rank test; 9 recordings from 2 rats. See also Figures S4 and S5 and Video S2.

Current Biology 29, 1–12, October 7, 2019 7

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

The Effect of Temporal Uncertainty on Tickling-Evoked Sensations We also made an attempt to test Aristotle’s [2] and Darwin’s [9] idea that predictability might underlie tickle suppression in our nose-poke-to-be-tickled paradigm. Specifically, we tested whether the temporal uncertainty of tickling affected the ticklishness as measured by the rate of evoked vocalizations. Instead of tickling directly after nose poking, we tickled the rats with variable delays up to several seconds (1.7 ± 0.1 s; mean ± SEM). We reasoned that this manipulation should markedly increase the temporal uncertainty of the delayed tickling events. In case temporal predictability determines ticklishness, we would expect increased vocalization rates for these delayed events. As shown in Figure S5, however, this was not the case. Vocalization rates were the same for immediate and delayed tickling, and there was no significant correlation between vocalization rate and delay. Hence, our data do not support a link between predictability and tickle effect. DISCUSSION Summary We observed striking differences in cortical and behavioral responses to self-touch and allo-touch or tickling. Specifically, we found a decrease of vocalization rate and cortical inhibition during self-touch as well as an increase of vocalization rate and cortical excitation during allo-touch and tickling. Our pharmacological experiments showed that GABAergic transmission is necessary for self-touch-induced suppression of allo-touchevoked vocalizations. Our self-initiated tickling paradigm confirmed the rewarding nature of tickling but also revealed a remarkable degree of ambivalence in tickling-associated behaviors. Ambivalence was evident in a large fraction of aborted trials, escape responses, freezing, and alarm calls. Finally, cortical recordings during self-initiated tickling interactions showed strong anticipatory responses to tickling in layer 5, but not in layer 4, of trunk somatosensory cortex.

Figure 6. Sensory Layer 4 Responses and Anticipatory Layer 5a Responses to Tickling (A) Top: color coding of behavioral events during a self-initiated tickling episode. Bottom: raw spike trace during a recording in a layer 4 of trunk somatosensory cortex during a self-initiated tickling episode is shown. (B) Same as (A) but for layer 5a recording. (C) Top: raster plot of layer 4 cell spikes (gray tick marks) and behavioral events (underlaid as color coded) aligned to the offset of freezing, sorted by duration of freezing. Bottom: PSTH of firing rate with the same alignment is shown. Data are binned into 250-ms width. (D) Same as (C) but for a layer 5a neuron. Note the difference in response onset between (C) (layer 4) and (D) (layer 5a). (E) Population PSTH of layer 4 responses aligned to the onset of self-initiated tickling. Averaged over 57 phases; 7 cells; 2 rats. Data are binned into 0.5-s width. (F) Same as (E) but for layer 5a recordings. Averaged over 55 phases; 17 cells; 2 rats. See also Figures S3 and S4.

8 Current Biology 29, 1–12, October 7, 2019

Behavioral and Neuronal Responses to Self-Touch, AlloTouch, and Tickling We studied allo-touch, tickling by the experimenter, and selftouch in the form of grooming or scratching. The grooming model of self-touch proved particularly advantageous, because of the spontaneous and frequent occurrence and the good body coverage. Allo-touch, tickling, and self-touch were associated with remarkably different vocalization behaviors. Although allotouch and tickling evoked vocalizations, self-touch was associated with decreased vocalization rates. We were impressed by the robust vocalization responses to allo-touch; in this context, it should be considered that these touch-evoked vocalizations were measured in a playful context, namely during tickling sessions. A fact that complicates the interpretation of our data is that we cannot verbally instruct animals and that we have no control over their mindset. Thus, experiments on humans that can be precisely instructed to perform specific forms of self-touch and that can be questioned on their subjective experience should complement our work. We also note that our conclusions rest on the validity of our ‘‘grooming-scratching model of self-touch.’’

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Our data indicate that self-touch evokes very different cortical responses compared to allo-touch or tickling, i.e., cortical inhibition versus excitation, respectively. The finding that allo-touch activates somatosensory cortex is in line both with the earliest observations on tactile responses in somatosensory cortex [18] and with subsequent work on social touch [19–21]. Also the strong excitatory effects of tickling on trunk cortex neurons and the tight correlation between allo-touch and tickling responses have been previously observed [8]. The inhibition evoked by self-touch is remarkably different from allo-touch and tickling responses. This suppression of somatosensory cortex activity by self-touch has been reported previously by imaging work [12]. Our data extend these earlier findings in six ways: (1) we show such inhibition on the cellular level; (2) we show that inhibition is the signature of very different forms of self-touch, i.e., grooming and scratching; and (3) our data indicate that there is a range of responses to self-touch, including some excitatory responses (Figures 1D–1F, 2D, and 2E). We wonder whether the excitatory responses are involved in mediating the sensation of self-touch; (4) we show that self-touch inhibition applies to most cortical layers; (5) we show that the inhibition evoked by self-touch is global, i.e., we observe robust inhibition in trunk cortex neurons in response to head grooming; and (6) we find that various forms of self-touch (trunk grooming, head grooming, and scratching) evoke responses of correlated strength (Figures 2D and 2E), whereas such self-touch responses are not (Figure 1D) or only very weakly (Figure 1E) correlated to allo-touch. Earlier imaging work suggested that the cerebellum might be suppressing responses to self-touch [12, 13]. The awareness of the ownership of the action, i.e., the sense of agency, is suggested to be mediated by the angular gyrus [22, 23]. The relatedness of self-touch responses might indicate a different source of such responses (cerebellar or angular gyrus inhibition?) from the allo-touch/tickling responses, which are again related to each other and might be both driven from the same source (i.e., afferents in the skin). With respect to self-tickle suppression, our most stunning observation is certainly the powerful suppression of vocalizations and cortical responses evoked by allo-touch and the suppression of vocalizations evoked by cortical microstimulation (Figure 3). This experiment was inspired by our physiological finding that inhibition evoked by self-touch is global and is by no means restricted to the body location contacted by self-touch (Figures 2 and S1). These results indicate that the somatosensory cortex does not compute a self- versus allo-touch distinction, because if this were true, allo-touch during self-touch should remain effective. Finally, we showed that intact GABAergic transmission in trunk cortex is necessary for a selftouch-induced suppression of allo-touch-evoked vocalizations. To our knowledge, this is the first time self-tickle suppression could be abolished by a neural manipulation. A potential caveat of this experiment is that blocking cortical inhibition is a drastic manipulation, which certainly increases cortical firing rates and might also result in neuronal oscillations. What adds to the significance of this observation is the specificity of our manipulation. Thus, we that blockade of GABAergic synapses in a few columns of trunk cortex is suffices to abolish the complex phenomenon of self-tickle suppression. This finding firmly establishes trunk somatosensory cortex as the site of self-tickle suppression,

much like our earlier work established that trunk somatosensory cortex is a neural substrate of the tickle response [8]. Our findings (self-touch-evoked inhibition, self-touch-evoked suppression of neural responses to allo-touch, self-touchevoked suppression of vocalizations elicited by allo-touch, and abolishment of self-tickle suppression by blockade of inhibition) are consistent with the global-inhibition model of self-tickle suppression. Such evidence does not amount to mechanistic certainty about the origin of self-tickle suppression, however. Further work in humans is needed to differentially test predictions of different models of self-tickle suppression. Self-Initiated Tickling and Ambivalence The self-initiated tickling paradigm—nose-poking in return for tickling interactions—confirmed earlier findings suggesting that tickling is rewarding to rats [16]. Unexpectedly, however, the rats also showed behavioral responses that are classically associated with aversive behavioral settings. In particular, the rats not only aborted a large fraction of nose poke attempts but also showed escape spurts, freezing, and emitted alarm calls. The post-nose-poke freezing is particularly puzzling, as freezing is a well-characterized post-threat-detection behavior [24]. We are convinced of the validity of our observation because rats expressed multiple aspects of freezing behavior, i.e., immobility, crouched body position, and an absence of vocalizations. This is not to say that the freezing observed in our self-initiated tickling paradigm is identical to predator or footshock-evoked freezing, which is often more marked and much more extended in time [17]. The post-nose-poke freezing behavior resembles stretch attend posture (SAP), a rodent behavior described as long stretching of the body while standing still or moving slowly forward [25]. SAP is classified as an ambivalent behavior reflecting an approach-avoidance tendency in social and non-social context [26]. Although SAP is observed under anxiogenic conditions and reduced by anxiolytic drugs [27], post-nose-poke freezing described in the current study was displayed in a playful context. Such a freezing behavior following self-initiation in animals has, to our knowledge, never been reported before. Why do rats show such ambivalent behavior? The occurrence of these behaviors related to aversive conditions is at first sight very odd, because the animal itself initiated the interaction by nose poking. We reckon that the occurrence of such behaviors can be understood in the relationship to play fighting. In the context of play fighting, escape spurts, freezing, and alarm calls are signals of defensive attitudes that can be understood by a play-mate. From our own personal experience of tickling children, this interpretation makes a lot of sense [28]. Children often struggle with the tickler, consistent with play fighting, and also cry for help [29]. As children also enjoy being tickled under such conditions, we suggest such behavior reflects ambivalent coexistence of reinforcement and fear or a ‘‘Nervenkitzel.’’ The self-initiated tickling paradigm also allowed us to vary the temporal uncertainty of tickle events (Figure S5). From our results, it is obvious that the temporal uncertainty of tickling events has no impact on the evoked vocalization rate. Thus, these data fail to support Aristotle’s [2] and Darwin’s [9] suggestion that unpredictability promotes the tickle sensation. Current Biology 29, 1–12, October 7, 2019 9

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

adopted during freezing was often interrupted just prior to tickling (see the raster plots in Figure 5E). In the self-initiated tickling paradigm, the animals knew that they would be tickled shortly. Under such circumstances, we observed a ramping up of layer 5 activity prior to tickling (Figures 6B, 6D, and 6F). Indeed, some layer 5a cells would discharge maximally prior to tickling onset (Figure 6B). Such anticipatory activity was not seen in layer 4 of trunk cortex (Figures 6A, 6C, and 6E). Thus, we suggest that anticipatory neuronal activity may arise from cortical top-down signals, which often target layer 1 and infragranular layers. Anticipatory vocalization and cortical response were a robust experimental finding also observed when we slowly approached the animals with the tickling hand (Figure S4). In this paradigm, however, the anticipatory cortical response seemed less layer specific and more ubiquitous. Anticipatory laughter is a curious feature of ticklishness and, similar to the ambivalent behaviors discussed above, we speculate that such vocalizations might be instrumental in engaging play-mates in play fighting, as they signal both positive affect and tickling expectation, thus inviting play attacks [29]. Models of Ticklish Experience Finally, we compare different models of ticklish experience. In Figure 7A, we present a model of ticklish experience based on our data from behavioral experiments and cortical recordings in rats. We refer to this scheme as global-inhibition-suppression model. The model outputs evoke vocalizations, which according to the data of Ishiyama and Brecht [8] are driven by layer 5 neurons in trunk somatosensory cortex. Three input sources modulate the vocalizations:

Figure 7. Models of Self-Tickle Suppression (A) The global-GABAergic-inhibition-suppression model predicts vocalization output based on our data from behavioral experiments, cortical recordings, and pharmacological experiments in rats. There is no computation of self versus other, and self-tickle suppression is achieved by global self-touchinduced inhibition. See text for details. (B) A ‘‘self-computation’’ suppression model adapted from Leavens and Bard [31]. The distinction of self versus other is at the heart of the model and mediates self-tickle suppression. See text for details.

Tickling and Anticipation Anticipatory laughter is a prominent yet little investigated behavioral feature of ticklish laughter [30]. An earlier study showed that anticipation of rewarding brain stimulation evokes vocalizations in rats [15]. Anticipatory vocalizations were evident also in the self-initiated tickling paradigm, where the silence that animals 10 Current Biology 29, 1–12, October 7, 2019

(1) Afferent excitation leads a specific, columnar activation of somatosensory cortex. Presumably, self-touch, allotouch, and tickling activate this pathway, which might also explain the correlation in cortical response strength between allo-touch and tickling. (2) The second input is GABAergic self-touch inhibition, which globally inhibits somatosensory cortex. Presumably, all forms of self-touch (head grooming, trunk grooming, scratching, etc.) activate this pathway, explaining the correlation in cortical response strength between different forms of self-touch. The neural origin of the self-touch inhibition signal is unknown. (3) The third input is an excitatory anticipation signal, which acts predominantly on layer 5A of somatosensory cortex. The neural origin of this anticipatory signal is unknown. The model generates vocalization in response to allo-touch, tickling, and anticipation of tickling. Self-touch, on the other hand, does not lead to vocalization, because, in this case, afferent excitation is overruled by global inhibition. The model also does not generate vocalization when self-touch and allotouch are co-applied, again because afferent excitation is overruled by global inhibition. Hence, the model operates without a formal computation of self versus other in the somatosensation. In Figure 7B, we show a classic ‘‘self-computation’’ model of ticklish experience; also, this model was derived from physiological observations. Self-tickle suppression is based on self versus other computation (adapted from Leavens and Bard) [31].

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Specifically, the model uses internally generated predictions to distinguish self from other, and in this computation, a ‘‘self’’ is at the heart of the model. We think this model is incompatible with the behavioral effects of the co-application of self- and allo-touch in rats. Given its computational structure, this model should distinguish self- from allo-touch and in consequence respond with ticklish sensation to allo-touch co-applied with self-touch, unlike what we observed in rats. It will be important to further test these models, because they suggest very different interpretations of behavior. For example, patients with schizophrenic symptoms [32] and healthy subjects woken from REM sleep [33] have been reported to be able to self-tickle. In light of a ‘‘self-computation’’ model of self-tickle suppression (Figure 7B), such effects might indicate a disruption of self-computation in schizophrenic patients or people woken from REM sleep. Should further work indicate, however, that there is no self-computation in tickling suppression as suggested in Figure 7A, the changes in these individuals will have a different basis. Thus, further work should test the different predictions of the global-inhibition-suppression model (Figure 6A) or the selfcomputation model (Figure 7B), for example, by co-applying self-touch and tickling in humans. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d d

d

d

KEY RESOURCES TABLE LEAD CONTACT AND MATERIALS AVAILABILITY EXPERIMENTAL MODEL AND SUBJECT DETAILS B Animals METHOD DETAILS B Operant conditioning B Behavioral recording and analyses B Electrophysiology and histology B Peri-self-touch tapping B Pharmacology B Slow approach B Microstimulation QUANTIFICATION AND STATISTICAL ANALYSIS B Motion tracking B Statistical analysis DATA AND CODE AVAILABILITY

SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j. cub.2019.07.085. A video abstract is available at https://doi.org/10.1016/j.cub.2019.07. 085#mmc5. ACKNOWLEDGMENTS This work was supported by Bernstein Center for Computational Neurosci€t zu Berlin, Germany, ‘‘Structural, neural ence Berlin, Humboldt-Universita and mental underpinnings of rat ticklishness’’ (Deutsche Forschungsgemeinschaft, Germany; project number: 393810148), and the Deutsche Forschungsgemeinschaft, Germany, Leibniz Prize. The authors thank U. Schneeweiß, E. Maier, A. Clemens, J. Sigl-Glo¨ckner, V. Bahr, F. Mielke, M. Kunert, A. Stern,

and J. Panksepp. Data are archived at the BCCN Berlin server and will be available for download upon request. AUTHOR CONTRIBUTIONS Conceptualization, S.I. and M.B.; Methodology, S.I. and M.B.; Investigation, S.I. and L.V.K.; Formal Analysis, S.I. and L.V.K.; Visualization, S.I.; Writing – Original Draft, S.I. and M.B.; Writing – Review & Editing, S.I., L.V.K., and M.B.; Supervision, M.B.; Funding Acquisition, S.I. and M.B. DECLARATION OF INTERESTS The authors declare no competing interests. Received: May 13, 2019 Revised: July 23, 2019 Accepted: July 30, 2019 Published: September 26, 2019 REFERENCES , A. (1897). The psychology of tickling, laughing, and the 1. Hall, G.S., and Allin comic. Am. J. Psychol. 9, 1–41. 2. Hett, W.S., and Rackham, H. (1936). Aristotle’s Problems, Volume 2 (Harvard University Press). 3. Panksepp, J., and Burgdorf, J. (1999). Laughing rats? Playful tickling arouses high frequency ultrasonic chirping in young rodents. In Toward a Science of Consciousness III, S.R. Hameroff, D. Chalmers, and A.W. Kaszniak, eds. (MIT Press), pp. 231–244. 4. Panksepp, J., and Burgdorf, J. (2000). 50-kHz chirping (laughter?) in response to conditioned and unconditioned tickle-induced reward in rats: effects of social housing and genetic variables. Behav. Brain Res. 115, 25–38. 5. Panksepp, J., and Burgdorf, J. (2003). ‘‘Laughing’’ rats and the evolutionary antecedents of human joy? Physiol. Behav. 79, 533–547. 6. Panksepp, J., Siviy, S., and Normansell, L. (1984). The psychobiology of play: theoretical and methodological perspectives. Neurosci. Biobehav. Rev. 8, 465–492. 7. Burgdorf, J., Wood, P.L., Kroes, R.A., Moskal, J.R., and Panksepp, J. (2007). Neurobiology of 50-kHz ultrasonic vocalizations in rats: electrode mapping, lesion, and pharmacology studies. Behav. Brain Res. 182, 274–283. 8. Ishiyama, S., and Brecht, M. (2016). Neural correlates of ticklishness in the rat somatosensory cortex. Science 354, 757–760. 9. Darwin, C. (1872). The Expressions of the Emotions in Man and Animals (John Murray). 10. Weiskrantz, L., Elliott, J., and Darlington, C. (1971). Preliminary observations on tickling oneself. Nature 230, 598–599. 11. Harris, C.R., and Christenfeld, N. (1999). Can a machine tickle? Psychon. Bull. Rev. 6, 504–510. 12. Blakemore, S.J., Wolpert, D.M., and Frith, C.D. (1998). Central cancellation of self-produced tickle sensation. Nat. Neurosci. 1, 635–640. 13. Blakemore, S.J., Wolpert, D., and Frith, C. (2000). Why can’t you tickle yourself? Neuroreport 11, R11–R16. 14. Berridge, K.C., Fentress, J.C., and Parr, H. (1987). Natural syntax rules control action sequence of rats. Behav. Brain Res. 23, 59–68. 15. Knutson, B., Burgdorf, J., and Panksepp, J. (1998). Anticipation of play elicits high-frequency ultrasonic vocalizations in young rats. J. Comp. Psychol. 112, 65–73. 16. Burgdorf, J., and Panksepp, J. (2001). Tickling induces reward in adolescent rats. Physiol. Behav. 72, 167–173. 17. Griffith, C.R. (1920). The behavior of white rats in the presence of cats. Psychobiology 2, 19–28.

Current Biology 29, 1–12, October 7, 2019 11

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

18. Mountcastle, V.B. (1957). Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J. Neurophysiol. 20, 408–434.

28. Ambrose, A. (1963). The age of onset of ambivalence in early infancy: indications from the study of laughing. J. Child Psychol. Psychiatry 4, 167–181.

19. Gazzola, V., Spezio, M.L., Etzel, J.A., Castelli, F., Adolphs, R., and Keysers, C. (2012). Primary somatosensory cortex discriminates affective significance in social touch. Proc. Natl. Acad. Sci. USA 109, E1657–E1666.

29. Harris, C.R. (2012). Tickling. In Encyclopedia of Human Behavior, Second Edition, V.S. Ramachandran, ed. (Academic Press), pp. 611–615.

20. Bobrov, E., Wolfe, J., Rao, R.P., and Brecht, M. (2014). The representation of social facial touch in rat barrel cortex. Curr. Biol. 24, 109–115.

30. Hoshikawa, T. (1991). Effects of attention and expectation on tickle sensation. Percept. Mot. Skills 72, 27–33.

21. Lenschow, C., and Brecht, M. (2015). Barrel cortex membrane potential dynamics in social touch. Neuron 85, 718–725.

31. Leavens, D.A., and Bard, K.A. (2016). Tickling. Curr. Biol. 26, R91–R93.

22. Farrer, C., Frey, S.H., Van Horn, J.D., Tunik, E., Turk, D., Inati, S., and Grafton, S.T. (2008). The angular gyrus computes action awareness representations. Cereb. Cortex 18, 254–261.

32. Blakemore, S.J., Smith, J., Steel, R., Johnstone, C.E., and Frith, C.D. (2000). The perception of self-produced sensory stimuli in patients with auditory hallucinations and passivity experiences: evidence for a breakdown in self-monitoring. Psychol. Med. 30, 1131–1139.

23. Chambon, V., Wenke, D., Fleming, S.M., Prinz, W., and Haggard, P. (2013). An online neural substrate for sense of agency. Cereb. Cortex 23, 1031– 1037. 24. Fanselow, M.S. (1994). Neural organization of the defensive behavior system responsible for fear. Psychon. Bull. Rev. 1, 429–438. 25. Grant, E.C., and Mackintosh, J.H. (1963). A comparison of the social postures of some common laboratory rodents. Behaviour 21, 246–259. 26. Kaesermann, H.P. (1986). Stretched attend posture, a non-social form of ambivalence, is sensitive to a conflict-reducing drug action. Psychopharmacology (Berl.) 89, 31–37. 27. Molewijk, H.E., van der Poel, A.M., and Olivier, B. (1995). The ambivalent behaviour ‘‘stretched approach posture’’ in the rat as a paradigm to characterize anxiolytic drugs. Psychopharmacology (Berl.) 121, 81–90.

12 Current Biology 29, 1–12, October 7, 2019

33. Blagrove, M., Blakemore, S.J., and Thayer, B.R. (2006). The ability to selftickle following rapid eye movement sleep dreaming. Conscious. Cogn. 15, 285–294. 34. Lausberg, H., and Sloetjes, H. (2009). Coding gestural behavior with the NEUROGES–ELAN system. Behav. Res. Methods 41, 841–849. 35. Rao, R.P., Mielke, F., Bobrov, E., and Brecht, M. (2014). Vocalizationwhisking coordination and multisensory integration of social signals in rat auditory cortex. eLife 3, e03185. 36. Brecht, M., and Sakmann, B. (2002). Dynamic representation of whisker deflection by synaptic potentials in spiny stellate and pyramidal cells in the barrels and septa of layer 4 rat somatosensory cortex. J. Physiol. 543, 49–70.

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Sigma-Aldrich

Cat# S106

This study

https://doi.org/10.17632/55xjvmc9pt.1

Janvier Labs

https://www.janvier-labs.com/

Neuralynx

https://neuralynx.com/software/cheetah

Avisoft RECORDER USGH

Avisoft Bioacoustics

https://www.avisoft.com/recorder.htm

MATLAB 2017b

MathWorks

https://www.mathworks.com/products/matlab.html

MClust 3.5

A. David Redish

http://redishlab.neuroscience.umn.edu/MClust/ MClust.html

Chemicals, Peptides, and Recombinant Proteins SR-95531 Deposited Data Source data for figures Experimental Models: Organisms/Strains Rat: RjOrl:LE Software and Algorithms Cheetah

Anaconda 3 (Python 3.6)

Anaconda

https://www.anaconda.com/download/

ELAN 5.1

[34]

https://tla.mpi.nl/tools/tla-tools/elan/

After Effects CC 2017

Adobe Systems

https://www.adobe.com/products/aftereffects.html

LEAD CONTACT AND MATERIALS AVAILABILITY This study did not generate new unique reagents. Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Michael Brecht ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS Animals Fourteen male juvenile Long-Evans rats were commercially acquired (3-week old at training onset) and individually caged in polycarbonate cages with wood chip bedding, maintained with a 12:12 h inverted light/dark cycle. Animals were allowed ad libitum access to food and water. Animals received 15-min handling and 15-min tickling session once a day for 2-4 weeks prior to experiments. All experimental procedures were performed according to German guidelines on animal welfare under supervision of local ethics committees in accordance to animal experimentation permit (Permit No. G0193/14). METHOD DETAILS Operant conditioning Rats received tickle stimuli in a test box (‘tickle box’) as previously described [8] for 15 min daily in the first training week. From the second training week, the test box was equipped with a column with a nose-poke hole (2.5 cm diameter; center of the hole 4 cm above the floor; infra-red LED and sensor located 5 mm from the opening of the hole, Figure 5A) and a beeper (resonance frequency, 1740 Hz). The training began with classical conditioning to associate the beep and tickling: the beep (0.5 s) was manually delivered by the experimenter before each tickling stimulus, 30 min daily for 3-5 days. The classical conditioning was followed by operant conditioning to associate the hole, beep and tickling for 2 weeks: the condition to activate the beep was gradually associated with the nose-poke i.e., standing in front of the hole; looking straight into the hole; touching the hole rim; poking the hole. At the end of the operant conditioning, beep was automatically delivered only when the rats poked the hole deep enough to break the LED beam. The latency between the beep and tickling was manually varied (1.77 ± 0.12 s, mean ± SEM). Nose-pokes that were not deep enough to break the LED beam were identified as failures. Behavioral recording and analyses Ultrasonic vocalizations (USVs) and videography (30 fps) were recorded and analyzed as previously described [8]. USVs were identified using a custom-written software. All USVs were 50 kHz calls, unless otherwise denoted. Video analyses were performed using ELAN 5.1 software [34]. Head grooming, trunk grooming, paw chewing and scratching were categorized as self-touch behaviors. The data for the self-touch behaviors include meta-analyses of previously-reported data [8]. Current Biology 29, 1–12.e1–e3, October 7, 2019 e1

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

Electrophysiology and histology Single and multi unit activity from the left trunk somatosensory cortex (1 mm posterior; 2-4 mm lateral from Bregma) was recorded using a Harlan 8 drive, Cheetah software, and DigitalLynx SX (Neuralynx) and analyzed as previously reported [35]. Spike response index was calculated as (FR – FRb) / (FR + FRb), where FR is the average firing rate during whole time of given phase and FRb is the average firing rate during all break phases i.e., time between heterospecific interaction phases. Recording tracks of interest were labeled with electrolytic lesions by applying a DC current (8 s, 8 mA, electrode tip negative). Injection pipettes in pharmacology experiments and in some instances also tetrodes were labeled with the fluorescent dye DiI (Thermo Fisher Scientific) for post hoc detection. At the end of experiments animals received an overdose of the anesthetic, were transcardially perfused with a pre-fixative solution followed by a 4% formaldehyde (PFA) solution and the brain was histologically processed. Brains were cut in 100 mm coronal sections and stained for cytochrome-oxidase activity as described previously [36]. Layers were assigned to recording sites based on the characteristic cytochrome-oxidase activity signal, which is strong in layer-4, weak in layer5A and intermediate in layer 5B. Peri-self-touch tapping Tickle-trained animals were placed in the test box. Rats were randomly tapped on the lateral trunk by index finger of the experimenter out of or peri-self-touch. To prevent self-touch from being interrupted by approaching hand, experimenter’s hand was kept in the box or slowly approached to the rats without being noticed once the rats started self-touch behaviors. We categorized series of taps (5-10 times) that were initiated within 600 ms after self-touch offset as ‘peri-self-touch tapping’. Self-touch offset was defined as detaching of the paws or the snout from the self-touching body part. Self-touch stopped at 46.0 ± 34.4 ms after onset of peri-self-touch tapping (mean ± SEM, 71 phases). Omnilateral trunk was tapped for the behavioral experiment (Figures 3A and 3B). To observe neuronal responses to peri-self-touch tapping, we recorded single and multi unit activity in the left trunk somatosensory cortex. Rats were tapped on the right lateral trunk out of or peri-grooming right trunk. Spike response indices were calculated as (FR – FRpre) / (FR + FRpre), where FR is the average firing rate during tapping phases and FRpre is the average firing rate during 3-to-1 s before tapping onset. Pharmacology Rats were tickle-trained for 7 days before being implanted with an aluminum head-fixation post on the interparietal bone. Animals were then habituated to head fixation up to 30 min twice a day for a week. Head fixation was followed by tapping interactions for 30 min. Craniotomy and durotomy were performed above the left trunk somatosensory cortex (1 mm posterior; 2-4 mm lateral from Bregma). To determine the precise injection site, receptive field of the trunk cortex was mapped for the right lateral trunk by using a tungsten electrode, and vascular landmarks were documented. The brain was protected by Kwik-cast Sealant (World Precision Instruments). From two days after craniotomy, rats were head-fixed and gabazine (SR-95531, Sigma-Aldrich; dissolved in saline) or saline as a control was microinjected in the target position of trunk somatosensory cortex (12.5-100 pmol; 250-500 nl; 1-2 min; 1.5 mm depth from pia; microinjector: Stoelting). Pipette was removed 1 min after completion of microinjection, and rats were transferred to the test box. Rats received out of or peri-groom tapping on the left and right lateral trunk. The following events were detected in the videography analysis: left/right grooming; left/right tapping; left/right peri-groom tapping. Ultrasonic vocalizations were analyzed. The gabazine/saline conditions were blinded to the experimenters until completion of video and vocalization analyses. Experimenters did not listen to the transposed rat vocalizations during experiments. The injection pipette was painted with DiI (Thermo Fisher Scientific) for the last experiments and injection depth was histologically confirmed. Slow approach To investigate neuronal response to tickling anticipation, we recorded activity of trunk somatosensory cortex while experimenter’s hand was slowly approaching the rat. Interaction paradigm consisted of the following: 10 s baseline; (10 s dorsal trunk tickling; 15 s break; 10 s slow approach; 10 s dorsal trunk tickling; 15 s break) x 5. Ultrasonic vocalization rates and trunk somatosensory neuron firing rates are aligned to the onset of slow approach phases, and normalized to the maximum rates in each phase. Time was normalized to the duration of slow approach phases (9.98 ± 0.08 s, mean ± SEM, 76 phases) i.e., 0 = start of approach; 1 = end of approach = start of dorsal trunk tickling. Correlation between approach time and normalized vocalization/spike rates were tested by Pearson’s correlation coefficients. Microstimulation Microstimulation was performed through a tetrode wire using Master-8-vp (AMPI) and A365 stimulus isolator (World Precision Instruments). Stimulation pulse: monopolar; tip negative; 20-200 mA; 100 Hz; 2 s duration. Microstimulations were applied out of or during self-touch.

e2 Current Biology 29, 1–12.e1–e3, October 7, 2019

Please cite this article in press as: Ishiyama et al., Behavioral and Cortical Correlates of Self-Suppression, Anticipation, and Ambivalence in Rat Tickling, Current Biology (2019), https://doi.org/10.1016/j.cub.2019.07.085

QUANTIFICATION AND STATISTICAL ANALYSIS Motion tracking Motion of the animals in the nose-poke experiments was analyzed by tracking an infra-red LED attached to the headstage (approx. 6 cm above the head) in the videography (1024 3 768 pixels; 6 mm lens; installed 150 cm above the floor; 30 fps). Motion tracking analysis was performed using Adobe After Effects CC 2017. X- and Y-pixel coordinates of the LED at each frame, from which coordinates in [cm] were calculated, were converted to Euclidean distance across two consecutive frames, and then to speed [cm/s]. Statistical analysis Normally distributed data (tested by Shapiro-Wilk test) are given as mean ± SEM and intergroup comparisons were performed with paired t test for paired data. Data with non-normal distribution are represented as median and interquartile range, and comparisons were evaluated with signed-rank test for paired data. Linear fitting was performed by using linear regression and Pearson’s correlation coefficients were represented as r. The levels of significance are indicated as *: p < 0.05; **: p < 0.01; ***: p < 0.001; n refers to sample size. Data were analyzed with using MATLAB 2017b and Python 3.6. DATA AND CODE AVAILABILITY Source data for figures in the paper are available (https://doi.org/10.17632/55xjvmc9pt.1).

Current Biology 29, 1–12.e1–e3, October 7, 2019 e3