Neural correlates of ticklishness in the rat somatosensory cortex

See allHide authors and affiliations

Science  11 Nov 2016:
Vol. 354, Issue 6313, pp. 757-760
DOI: 10.1126/science.aah5114

Resolving a ticklish problem

What is the neural correlate of ticklishness? When Ishiyama and Brecht tickled rats, the animals produced noises and other joyful responses. During the tickling, the authors observed nerve cell activity in deep layers of the somatosensory cortex corresponding to the animals' trunks. Furthermore, microstimulation of this brain region evoked the same behavior. Just as in humans, mood could modulate this neuronal activity. Anxiety-inducing situations suppressed the cells' firing, and the animal could no longer be tickled.

Science, this issue p. 757


Rats emit ultrasonic vocalizations in response to tickling by humans. Tickling is rewarding through dopaminergic mechanisms, but the function and neural correlates of ticklishness are unknown. We confirmed that tickling of rats evoked vocalizations, approach, and unsolicited jumps (Freudensprünge). Recordings in the trunk region of the rat somatosensory cortex showed intense tickling-evoked activity in most neurons, whereas a minority of cells were suppressed by tickling. Tickling responses predicted nontactile neural responses to play behaviors, which suggests a neuronal link between tickling and play. Anxiogenic conditions suppressed tickling-evoked vocalizations and trunk cortex activity. Deep-layer trunk cortex neurons discharged during vocalizations, and deep-layer microstimulation evoked vocalizations. Our findings provide evidence for deep-layer trunk cortex activity as a neural correlate of ticklishness.

Tickling sensations can be differentiated into laughter-inducing “gargalesis” and non–laughter-inducing light touch, “knismesis” (1). The former is a peculiar, often funny form of social touch, which has been discussed for more than two millennia (2, 3). Still, major questions remain unanswered: Why does tickling induce laughter? Why are tickling effects so mood-dependent (4)? Why do body parts differ in ticklishness (1)? Why can’t we tickle ourselves (2)? Is ticklish laughter different from humorous laughter (46)? To address such questions, we need a better understanding of the neural correlates of ticklishness. We took advantage of groundbreaking advances that provided evidence for tickling-evoked 50-kHz vocalizations and primitive forms of joy in rats (7, 8). On the basis of this work, we measured “rat ticklishness” in our work as the propensity to “laugh” (emit 50-kHz calls) upon being tickled. We focused on the somatosensory cortex because it is the largest tactile neural representation in mammals, because human imaging studies suggested this candidate region (9, 10), and because work on somatosensory afferents provided no conclusive evidence for dedicated peripheral mechanisms of tickle.

We tickled and gently touched rats on different body parts (Fig. 1A, ventral tickling) and observed a variety of ultrasonic vocalizations (USVs; Fig. 1B), in particular during tickling (movie S1). Rat 50-kHz vocalizations indicate positive emotional valence (11). Consistent with earlier claims that tickling is rewarding through the dopaminergic system (7, 12, 13), rats rapidly approached the tickling hand, and tickling induced unsolicited jumps accompanied by 50-kHz USVs (Freudensprünge, “joy jumps”; movie S2), which can be seen in joyful subjects in various mammalian species (1416). We visually categorized spectrograms of an extensive set of USVs (34,140 calls) into modulated, trill, combined, and miscellaneous call types (Fig. 1B and fig. S1). Both tickling and gentle touch evoked USVs (Fig. 1C), but tickling induced more USVs than did gentle touch (ventral tickling, 4.45 ± 0.28 Hz; ventral gentle touch, 2.58 ± 0.21 Hz; n = 16, P < 0.001; mean ± SEM, paired t test). Rats seemed to warm up to tickling and vocalized less before the initial interaction than during breaks between interaction episodes (Pre versus Break; Fig. 1, C and D). Tickling the ventral trunk evoked the largest number of USVs (Fig. 1D) and the largest fraction of combined USVs (Fig. 1E). Play behavior (rat chasing experimenter’s hand; Fig. 1F) also evoked USVs (Fig. 1, G and H, and movie S3) (17). Consistent with our sense that rats experienced the experimental setting as emotionally positive, we did not observe 22-kHz alarm calls.

Fig. 1 Tickling and play behavior (chasing the experimenter’s hand) evoke ultrasonic vocalizations.

(A) Tickling of the ventral trunk of a rat. (B) Spectrograms of ultrasonic vocalizations (USVs). USVs were categorized into modulated (left), trill (middle), and combined (right) call types. (C) USVs during tickling and touch. Raster plots and beige boxes indicate USV onsets and interaction phases, respectively. (D) USV rate during different phases (n = 16 recordings from four animals). Data are means ± SEM. P value refers to analysis of variance (ANOVA); pairwise comparisons are denoted as ***P < 0.001 and *P = 0.014 (paired t test). (E) Fraction of USV types falling into the different categories [the same data from (D)] during tail tickling (1869 USVs) and ventral tickling (3380 USVs; combined, ***P < 0.001, Fisher exact test). (F) Play behavior: a rat chasing the experimenter’s hand. (G) USVs during chasing hand. (H) Fraction of USV types during chasing hand [3181 USVs; 15 recordings from three animals; colors as in (E)].

We simultaneously performed single-unit recordings in the trunk region of the somatosensory cortex (Fig. 2A). We obtained high-quality recordings of neuronal responses elicited by tickling and gentle touch (Fig. 2B and fig. S2A). Similar to USVs, activity in the trunk cortex was lower before initial tickling (Fig. 2, B and C, Pre) than during the short breaks between interactions (Fig. 2, B and C, Break). Remarkably, neuronal responses were also observed during hand-chasing phases, when rats were not touched by the experimenter (Fig. 2B and fig. S2E). Most cells increased their firing rate during trunk tickling, trunk gentle touch, and chasing hand (~77%, ~67%, and ~80% of the cells showed higher firing rates during interaction than during break, respectively; fig. S2B, top, and C to E), whereas a minority of cells were suppressed during interaction phases (fig. S2B, bottom, and C to E). Similar to USVs, neuronal firing rates increased more during tickling than during gentle touch on the trunk (Fig. 2D). As in the cells shown in fig. S2B, responses to tickling predicted play responses (chasing hand) across the population (Fig. 2E), which suggests a neural link between tickling and play behavior.

Fig. 2 Tickling and play behavior (chasing hand) modulate firing rate in trunk somatosensory cortex neurons.

(A) Left: Cytochrome oxidase stain of a coronal trunk somatosensory cortex section (scale bar, 200 μm; D, dorsal; L, lateral; WM, white matter). Right: Histology [black curves, layers; red ovals, lesions; dashed line, tetrode track; cross, recording site of (B)]. (B) Histogram of firing rate in a layer 5a cell during tickling, touch, and play (beige boxes; Pre, pre-interaction). Data were binned into 3-s intervals. (C) Firing rates in pre-interaction (Pre) versus in break (Wilcoxon signed-rank test). (D) Firing rates during trunk gentle touch versus trunk tickling (Wilcoxon signed-rank test). (E) Population data indicating a correlation between trunk tickling and chasing hand response indices (ρ, Pearson linear correlation coefficient; data fitted with line; n denotes number of cells).

Anxiogenic conditions suppress tickling-evoked USVs in rats (7). To test whether neuronal responses to tickling are also modulated by such conditions, we tickled rats in both control (Fig. 3A, left) and anxiogenic settings, such as under bright illumination and on an elevated platform (Fig. 3A, right). Tickling-evoked USVs were significantly suppressed in the anxiogenic condition and recovered in control conditions (Fig. 3, B and C). Similarly, anxiogenic conditions suppressed neuronal response to tickling (Fig. 3D) and inverted the sign of response index to tickling (Fig. 3E).

Fig. 3 Anxiogenic suppression of USVs and neuronal responses to tickling in trunk somatosensory cortex.

(A) Control (left) and anxiogenic condition (right). (B) USVs in control and anxiogenic conditions. Black boxes, interaction phases; raster plots, USVs; D, dorsal tickling; V, ventral tickling; Dg, dorsal gentle touch; Vg, ventral gentle touch; T, tail tickling; Ch, chasing hand; Misc., miscellaneous. (C) USV rate during trunk tickling under control versus anxiogenic conditions; n = 6 recordings from four animals. Data are means ± SEM (paired t test). (D) Peristimulus time histograms (PSTHs) of firing in a cell during dorsal tickling (beige boxes) under control and anxiogenic conditions. (E) Response indices of cells for trunk tickling under control versus anxiogenic conditions (n = 27 cells, 6 recordings from four animals). Data are means ± SEM (paired t test).

Our recordings revealed that USVs and neuronal activity in the trunk cortex are modulated in a similar way by tickling and anxiogenic conditions. We wondered whether tickling-evoked USVs and neuronal responses to tickling are causally linked. We therefore aligned neuronal firing to the onsets of USVs (Fig. 4, A and B). To avoid a confounding effect of the coactivation of the trunk cortex and USVs by tickling and touch, we restricted this analysis to break periods. The activity of trunk somatosensory neurons was correlated with USV emissions: Neurons increased their firing rate before and during USV emissions (Fig. 4, C and D, Before versus On). This effect was strongest for combined USVs (Fig. 4E). Furthermore, the effect was more prominent in layers 4 and 5a than in the superficial layers (Fig. 4F). To test whether firing of somatosensory neurons causes USVs, we microstimulated the trunk cortex (Fig. 4G). Although rats had no interaction with the experimenter, they emitted USVs (Fig. 4H, top). Threshold amplitudes to trigger USVs varied between 50 and 300 μA. USVs were locked to microstimulation onset and were evoked after short latencies of 50 to 100 ms (Fig. 4H, bottom). When microstimulation was directly preceded by tickling, more USVs were evoked and current thresholds were lower. Microstimulation in the deep layers, but not in the superficial layers, evoked USVs (Fig. 4I).

Fig. 4 Neuronal activity in deep layers of trunk somatosensory cortex is associated with USVs, and deep-layer microstimulation evokes USVs.

(A) Left: Cytochrome oxidase stain of a coronal trunk somatosensory cortex section (scale bar, 200 μm; D, dorsal; L, lateral; WM, white matter). Right: Histology [black curves, layers; red ovals, lesions; dashed line, tetrode track; cross, recording site of (B)]. (B) PSTH of firing rate in a layer 5a cell aligned to USV onsets (in break phases with no previous USV onset within 500 ms). (C) PSTH of firing rate aligned to the onset of USV (n denotes number of cells). (D) Firing rate before USV [–300 to –200 ms from USV onset, Before in (C)] versus firing rate upon USV [0 to 100 ms from USV onset, On in (C)]. The same population data from (C) are plotted for each cell (Wilcoxon signed-rank test). (E) Response indices for each USV type, using the same data from (C) and (D). Mod., modulated; Comb., combined. Data are means ± SEM. P value refers to one-way ANOVA; pairwise comparisons are denoted as **P = 0.007 and *P = 0.010 (paired t test). (F) Average response indices in different cortical layers. Numbers of cells are shown in parentheses. Data are means ± SEM. P value refers to one-way ANOVA on ranks; pairwise comparisons are denoted as ***P < 0.001 and **P = 0.003 (t test). (G) Microstimulation in trunk somatosensory cortex. (H) Top: USVs during microstimulation (beige boxes; 200 μA, 100 Hz, 2 s). Raster plots indicate USV onsets. Bottom: Average PSTH of USV rate aligned to stimulation onsets (averaged over 278 stimulations that evoked USVs at 18 sites from three animals; data were binned into 50-ms intervals). (I) Number of sites in different layers of trunk somatosensory cortex, where microstimulation did (“success,” black) or did not (“failure,” gray) evoke USVs. P value refers to χ2 statistics.

Our findings confirm key conclusions of Panksepp and Burgdorf (8): Rats vocalize during tickling in a mood-dependent fashion. The increase of vocalizations after initial tickling (Fig. 1, C and D) and anxiogenic suppression of tickling-evoked calls (Fig. 3, B and C) support Darwin’s idea that “the mind must be in a pleasurable condition” for ticklish laughter (4). Tickling-evoked calls are not simple reflexes in response to touch. Rats rarely emit combined calls during social facial touch with conspecifics (18, 19). Rats emitted combined calls preferentially during tickling on the belly, the most ticklish body part (as assessed from calling rate). Combined calls might be a relatively tickle-specific vocalization in rats. Remarkably, similar call types have been described during conspecific play behavior (20). The numerous similarities between rat and human ticklishness, such as tickling-evoked vocalizations and anxiogenic modulation, suggest that tickling is a very old and conserved form of social physicality.

Peripheral mechanisms of pleasurable touch were first studied by Zotterman in cats and suggested that knismesis is carried in part by pain fibers (21). C-fibers, unmyelinated afferents, are putatively involved in pleasurable touch in rodents (22). Central mechanisms of tickling were investigated by functional magnetic resonance imaging (fMRI) in human brains (9); that study, which used tickling stimuli evoking knismesis and observed somatosensory cortex activation, suggested that self-tickle suppression might be mediated by the cerebellum. Recent human fMRI identified activation of the lateral hypothalamus, parietal operculum, amygdala, cerebellum, and somatosensory cortex by ticklish laughter (10).

Four of our results localize tickle processing to the somatosensory cortex: (i) We found that tickling can evoke intense neuronal activity in the somatosensory cortex (Fig. 2). Moreover, play behavior, which induces anticipatory vocalizations in rats (Fig. 1, F to H) (17) and humans (23), evoked neuronal activity similar to the activity evoked by tickling (fig. S2E). (ii) We observed “mood-dependent” alteration of activity in the trunk somatosensory cortex, specifically an activity increase after tickling phases (Fig. 2, B and C), anxiogenic suppression of responses (Fig. 3, D and E), and a reduction of microstimulation thresholds for evoking calls after tickling. Such “mood-dependent” modulation of the somatosensory cortex is unexpected, as it is nontactile and there is little evidence to date for mood effects in other cortical areas. (iii) The strong call-related activation of the trunk somatosensory cortex points to an involvement in tickling-evoked vocalizations. Call-related firing in the somatosensory cortex is much stronger than call-evoked activity in the auditory cortex (18). (iv) Microstimulation-evoked vocalizations suggest that deep-layer but not superficial-layer cortical activity is sufficient to trigger vocalizations. The short latencies of evoked calls indicate few intervening processing steps between trunk activity and calls. Electrical stimulation in various brain areas is known to evoke laughter with or without mirth (2427), but stimulation-evoked ticklish laughter has not been reported so far, and our results might be different from previously reported laughter evoked by brain stimulation.

In line with lesion evidence, our observations suggest a neural link among tickling, play, and the somatosensory cortex (28). Other findings have implicated the somatosensory cortex in social information processing (2931). The observation that the somatosensory cortex is involved in the generation of tickling responses suggests that this area might be more closely involved in emotional processing than previously thought. Identification of the neural correlates of ticklishness will allow us to frame questions about tickling in neural terms and thus help us to understand this mysterious sensation.

Supplementary Materials

Materials and Methods

Figs. S1 and S2

Movies S1 to S3

References (3234)

References and Notes

  1. Acknowledgments: Supported by BCCN Berlin, Humboldt-Universität zu Berlin, SFB665, and the Deutsche Forschungsgemeinschaft Leibniz Prize. We thank V. Bahr, T. Balmer, A. Clemens, R. de Filippo, J. Diederichs, C. Ebbesen, K. Hartmann, M. Kunert, C. Lenschow, F. Mielke, W. Muñoz-Miranda, A. Neukirchner, C. Posey, R. Rao, U. Schneeweiß, and A. Stern. All of the data are archived at the BCCN Berlin server and will be available for download upon request.
View Abstract

Navigate This Article