Skip to main content
NCBI home page
Journal of Neurophysiology logoLink to Journal of Neurophysiology
. 2016 Jan 28;115(3):1703–1712. doi: 10.1152/jn.00460.2015

The tactile motion aftereffect suggests an intensive code for speed in neurons sensitive to both speed and direction of motion

S McIntyre 1,2,3,4,✉, I Birznieks 2,4,5, R M Vickery 2,5, A O Holcombe 1, T Seizova-Cajic 3
PMCID: PMC4808137  PMID: 26823511

Abstract

Neurophysiological studies in primates have found that direction-sensitive neurons in the primary somatosensory cortex (SI) generally increase their response rate with increasing speed of object motion across the skin and show little evidence of speed tuning. We employed psychophysics to determine whether human perception of motion direction could be explained by features of such neurons and whether evidence can be found for a speed-tuned process. After adaptation to motion across the skin, a subsequently presented dynamic test stimulus yields an impression of motion in the opposite direction. We measured the strength of this tactile motion aftereffect (tMAE) induced with different combinations of adapting and test speeds. Distal-to-proximal or proximal-to-distal adapting motion was applied to participants' index fingers using a tactile array, after which participants reported the perceived direction of a bidirectional test stimulus. An intensive code for speed, like that observed in SI neurons, predicts greater adaptation (and a stronger tMAE) the faster the adapting speed, regardless of the test speed. In contrast, speed tuning of direction-sensitive neurons predicts the greatest tMAE when the adapting and test stimuli have matching speeds. We found that the strength of the tMAE increased monotonically with adapting speed, regardless of the test speed, showing no evidence of speed tuning. Our data are consistent with neurophysiological findings that suggest an intensive code for speed along the motion processing pathways comprising neurons sensitive both to speed and direction of motion.

Keywords: touch, human, psychophysics, adaptation, motion

the ability of neurons in the primary somatosensory cortex (SI) to discriminate direction of motion across the skin usually increases with stimulus speed: the firing rates in neurons' preferred direction are greater than those in nonpreferred directions, and this difference tends to be greater with higher speeds of motion (Ruiz et al. 1995; Pei et al. 2010). SI responses to speed show little evidence of speed tuning, i.e., that different neural populations respond preferentially to particular speeds (Pei et al. 2010; Dépeault et al. 2013; although Pei et al. found that a minority of neurons were speed tuned). This is somewhat surprising, given that subjects are able to discriminate speed well (Dépeault et al. 2008) and that speed tuning is observed in the visual system (reviewed by Bradley and Goyal 2008), with which tactile motion shares many properties (Pack and Bensmaia 2015).

It is not clear how speed is coded in neural populations underlying perception of motion direction. We consider two possibilities. The first is that speed is coded simply by an increase in the firing rates of directionally tuned neurons. For example, slow speed in the distal direction would preferentially activate distal-selective neurons, and higher speed in the same direction would yield higher firing rates in those same neurons. We will refer to this possibility as the intensive code for speed (Fig. 1, left). Alternatively, different direction-sensitive neural populations may respond preferentially to particular speeds, and we will refer to this possibility as speed tuning (Fig. 1, right). In the case of speed tuning, a particular speed would activate a particular set of direction selective neurons, determined by the bandwidth of neuronal speed tuning.

Fig. 1.

Fig. 1.

Open in a new tab

Two models of speed coding in direction-selective tactile motion neurons. Solid lines show the population response before adaptation, and dotted lines the responsiveness of neurons after adaptation. Left: intensive code for speed: all neurons sensitive to a particular direction of motion within their receptive field respond to all speeds, but they respond more to faster speeds. Adaptation to fast motion would therefore adapt all neurons to a greater extent than adaptation using slow motion, as illustrated at bottom. Right: speed tuning: different subpopulations of direction-sensitive neurons respond optimally to different speeds. We illustrate the simplest case with two speed channels, slow and fast, with limited overlap. Adaptation to fast motion would adapt the fast channel but not the slow, and vice versa, as illustrated at bottom.

In the present study we utilize an adaptation paradigm to attempt to distinguish an intensive code for speed from speed tuning. This paradigm relies on the fact that tactile motion perception is subject to adaptation, such that sustained exposure to an adapting motion stimulus results in a change in the perception of a subsequently felt test stimulus. Historically, studies of adaptation to tactile motion yielded inconsistent effects (Thalman 1922; Hazlewood 1971; Planetta and Servos 2008), but we used a dynamic test stimulus, which has recently been shown (Watanabe et al. 2007; Konkle et al. 2009; Kuroki et al. 2012; McIntyre et al. 2014) to reliably elicit the tactile motion aftereffect (tMAE).

In this dynamic test version of the tMAE, exposure to motion across the skin results in a consistent bias to perceive the test stimulus as moving in the opposite direction to the adapting motion (Watanabe et al. 2007; Konkle et al. 2009; Kuroki et al. 2012; McIntyre et al. 2014). The existence of the tMAE is consistent with perceived direction being determined by the population response of speed encoding direction-sensitive neurons. Perceptual change following adaptation is explained by the shift in the population response profile (see Mather and Moulden 1980; Schwartz et al. 2007; Clifford 2014). The strength of perceptual aftereffects reflects the strength with which the adapting stimulus drives the neurons (Barlow and Hill 1963; Lundström and Johansson 1986; Harris et al. 2000; Clifford 2002; Schwartz et al. 2007).

Were speed coded intensively in direction-sensitive neurons, a fast adapting stimulus would produce strong neural responses, leading to strong adaptation affecting perception of movement direction. High speeds should therefore result in a tMAE regardless of the speed used for the test stimulus. The strength of the tMAE will be determined solely by the adapting speed (and not by the test speed); this possibility is illustrated in Fig. 2, left.

Fig. 2.

Fig. 2.

Open in a new tab

The predicted perceptual effects of adaptation for an intensive code for speed, speed tuning, and a combined code. Tactile motion aftereffect (tMAE), a perceptual response, is plotted as a function of adapting speed, on arbitrary scales. Lower values on the y-axis indicate a stronger aftereffect, reflecting a negative tMAE (opposite to the adapting motion). The effects of low (grey lines) and high (black lines) speeds of the test stimulus are shown. Left: an intensive code for speed produces an aftereffect that increases in strength monotonically with adapting speed and is the same for both speeds of the test stimulus. Middle: speed tuning results in the strongest aftereffect occurring when there is a correspondence between the speeds of the adapting and test stimuli; therefore, the low-speed test produces its strongest aftereffect with a lower adapting speed than does the high-speed test. Right: an intensive code for speed and speed tuning may be present at different stages of tactile motion processing, and both may be affected by adaptation. In this case, the aftereffect would reflect both mechanisms.

If speed is coded by speed-tuned neurons rather than intensively coded, an adapting stimulus of a given speed will produce the greatest neural responses (and hence adaptation) in the neurons most sensitive to that speed and less or no response in other neurons tuned to the same direction but to different speeds (depending on their respective bandwidths). This should result in a tMAE that is stronger when the adapting and test speed are similar, compared with when the test speed is much slower or faster than the adapting speed (Fig. 2, middle).

A third possibility is that both an intensive code for speed and speed tuning are present, with predicted results shown in Fig. 2, right.

Adapting and test stimuli in our study consisted of apparent motion created by a two-dimensional 6 × 24 pin array in a modified Optacon. Motion adaptation was induced by resting a finger on the array while moving object was simulated in each column of the array (see Fig. 3). To elicit the tMAE, we used a direction-balanced test stimulus containing motion in opposite directions, spatially interleaved: half of the pin array's columns moved in the same direction as the adapting stimulus, and the other half moved in the opposite direction.

Fig. 3.

Fig. 3.

Open in a new tab

Schematic representation of the tactile array and the activation patterns used in the 3-second adaptation experiment. A: a filled circle indicates an active (vibrating) pin. Left: the adapting stimulus: Pins in each column were successively activated to produce apparent motion in either a proximal-to-distal direction or vice versa. When activation reached the edge of the array, it started again at the other edge (illustrated in column 3). The pins active at any given time were spread across the array, as only one pin was active in each section (dashed outlines) at the start of stimulation. The specific row to be initially activated within each section was randomly determined. Right: the 2-component test stimulus: as in the adapting stimulus, pins were successively activated in each column, but here, half the columns stepped in one direction and the other half stepped in the other direction. B: the 1st 3 pins in 1 column are shown in the temporal sequence used to produce motion at 45 mm/s. C: each trial consisted of 3 s of the adapting stimulus followed by 3 s of the 2-component test stimulus. Additional test trials with no immediately preceding adapting stimulus (shown here at the 3rd stimulus presentation) were inserted in random locations in the series.

We used a variety of adapting speeds and conducted two experiments to test a total of four test speeds across the two experiments. The two experiments investigated the effects of adaptation at different time scales. The experiments are labeled the “3-s adaptation experiment” and the “10-s adaptation experiment.” In both experiments, tactile apparent motion of the vibrotactile array was applied to the finger pad of the index finger. Participants adapted to motion in the distal-to-proximal or the proximal-to-distal direction and judged the direction of motion (as a forced choice) of a subsequent test stimulus comprising simultaneous motion in both distal and proximal directions. The strength of the aftereffect was quantified for each adapt-test speed combination by the direction index (DI): the proportion of responses indicating that the test stimulus moved in the proximal direction subtracted from the proportion of responses indicating that it moved in the distal direction.

In summary, we used unidirectional adapting motion to “adapt out” one of the two directions of motion present in our bidirectional test stimulus. The expected consequence was perception of motion in the opposite direction, tMAE (Watanabe et al. 2007; McIntyre et al. 2014). Importantly, the strength of this direction percept would depend on the relative strength of the two direction signals, which are in turn affected by adapting and test speeds, in a manner reflecting the nature of neural coding in speed and direction-sensitive neurons.

METHODS

Participants.

Participants provided informed written consent, and the protocol was approved by the ethics committee of the University of New South Wales in accordance with the Declaration of Helsinki (2008 version). Naïve participants were compensated for their time. Eight volunteered for the 3-s adaptation experiment, including six naïve and two authors (aged 18–47, 4 female). Nine volunteers participated in the prolonged adaptation experiment (including 3 from the first experiment), although two were excluded because an insufficient number of responses were recorded (see Prolonged adaptation experiment). Of the remaining seven participants, five were naïve, and two were authors (aged 18–38, 3 female).

Apparatus.

Stimulation was applied using a 6 × 24 array of small pins (diameter 250 μm) that vibrated at up to 230 Hz (Optacon 1C; Telesensory Systems, Palo Alto, CA). The tactile array covered a total area of 27.2 × 11.4 mm, with rows 1.18 mm apart and columns 2.29 mm apart. Each protraction of a pin was driven by voltage changes of a fixed temporal profile, with a protraction time of ∼2 ms. The duration and amplitude (maximum of 65 μm, less when damped by the contacting finger; Bliss 1970) of the protraction were independent of the frequency of activation.

Participants rested their right arm on a cushion with their hand palm up. The index finger was fixed by sticking its dorsal side to a finger support using Play-Doh (Hasbro). The tactile array, mounted on a counterweighted metal arm, was lowered onto the distal region of the right index finger pad. The counterweight was adjusted so that the normal force exerted on the finger by the tactile apparatus was ∼0.3 N. This force setting was determined to be comfortable and provide reasonable contact with the tactile array, while avoiding excessive damping of the vibration. Participants wore earplugs under headphones through which white noise played to mask the sound of the vibrating pins. Participants responded by pressing buttons with the left hand.

Stimulus design.

The adapting and test stimuli for the 3-s adaptation experiment are illustrated in Fig. 3. Neighboring pins were successively activated (causing them to vibrate at 230 Hz) to create apparent motion. The adapting stimulus used all 24 rows and the center four columns. Pins in successive positions of the tactile array were activated in a proximal-to-distal direction or vice versa. The test stimulus was identical to the adapting stimulus except that the two odd columns were used to create motion in one direction (randomly determined for each trial) and the two even columns in the other. This produced a balanced dynamic test stimulus with equal motion signals in each direction, providing a sensitive test of the direction of any motion aftereffect (McIntyre et al. 2014).

The speed of apparent motion was set by the temporal delay [interstimulus onset interval (ISOI)] between activating one pin and the next (see Fig. 3B). There was no delay between the offset of one pin and the onset of the next, meaning that each activated pin vibrated for the entire ISOI. Rather than simulate a moving bar (Gardner and Sklar 1994; Essick et al. 1996; Watanabe et al. 2007; Konkle et al. 2009), we chose to spatially offset the activated columns from their active neighbors, as if the stimulus was a set of moving dots. This avoided the discontinuous “jump” associated with repetitive “sweeps” of a bar stimulus when it reaches the end position of the array and reappears at the other end, allowing us to continuously loop the stimulus without interruption. This stimulus gave a continuous impression of motion spread across the array, in contrast to a bar which presents a discrete position that might be tracked, potentially involving an attentional, position-based motion system (e.g., Cavanagh 1992) in addition to lower level motion processing.

Three-second adaptation experiment.

Four adapting speeds (19, 27, 45, and 136 mm/s) were used, which corresponded to the following ISOIs between neighboring pins: 69.6, 43.5, 26.1, and 8.7 ms. The inverse of the ISOI is the step frequency (16, 23, 38, and 115 steps/s). The speed of the test stimulus was 19 mm/s for half of the trials of each adapting condition and 136 mm/s for the other half. To maximize the possibility of detecting even very broad speed tuning, a wide range was used, covering the range of speeds typical of human spontaneous movements during surface exploration (Smith et al. 2002; Yoshioka et al. 2011). These speeds are suited for direction discrimination (Essick et al. 1991; Gardner and Sklar 1994) and for judging surface roughness (Lederman 1974 1983; Meftah et al. 2000). Each combination of adapting speed (19, 27, 45, and 136 mm/s and none) and test speed (19 an 136 mm/s) was repeated 48 times per subject in a random order with the adapting direction distal in half the trials and proximal in the other half. The trials in which the test stimulus had no preceding adapting stimulus were inserted in random locations in the series to break the predictability of the adapt-test sequence, and serve as a test of the direction perceived when the preceding trial had no net motion direction. Intervals of 1 s in which the pins did not vibrate separated all the stimulus presentations.

Procedure.

An individual trial consisted of 3 s of adaptation in one direction, followed by 3 s of the two-component test stimulus. The test stimulus was intended to reveal the direction of any aftereffect. Participants did six 10-min sessions, one trial after another (see Fig. 3C), in which each combination of test speed, adapting speed (including no adaptation), and adapting direction (proximal, distal) was tested, in a random order. For each participant, each combination was repeated 48 times over the whole experiment (24 times for each adapting direction). Participants reported the perceived direction of each 3-s stimulus (both the adapting and test stimuli) as either “distal” or “proximal” by pressing a button. Naïve participants were not aware of the nature of, or difference between, the adapting and test stimuli.

Ten-second adaptation experiment.

This experiment examined the effect of longer adapting stimuli and different speeds using a similar procedure to the 3-s adaptation experiment. The main methodological differences wereas follows: the duration of the adapting and test stimuli (10 s of continuous presentation each); all trials with a given combination of adapting and test stimuli were blocked as a single testing session (10 repeats); a different set of adapting speeds (23, 34, 45, 68, and 136 mm/s) and test speeds (34 and 68 mm/s) was used; and all six columns of the Optacon were used. This protocol allowed for the possibility that adaptation might build up over time during a session. Participants were also tested in a “no adaptation” condition, in which the test stimulus was presented in the same way as in the adaptation conditions, but there were gaps of no stimulation where adaptation had occurred in the other conditions. Participants had a break of at least an hour after each testing session before the subsequent session began (using a different adapting stimulus).

To ensure a continuous record of perception of the stimulus, participants were required to provide a response at least every 2 s throughout the 10-s stimulus period, regardless of whether the percept had changed. In addition to the “proximal” and “distal” responses available in the main experiment, participants also had the option of responding that the stimulus direction was “unclear.” This was so we could verify that participants perceived motion when feeling the bidirectional test. Subject responses were interpolated at 0.1-s intervals between button presses, with the first percept considered to begin with the first button press. The proportion of the total duration spent in each perceptual state (distal, proximal, and unclear) was calculated for each stimulus presentation. Trials in which 4 s or more elapsed with no response to a stimulus were discarded. In addition, two participants were excluded from data analysis because on more than 5% of trials insufficient button presses, as defined above, were recorded.

Data analysis.

The DI was calculated for each adapt-test speed combination. The DI is the proportion of trials the stimulus was perceived moving in the proximal direction, subtracted from the proportion of time it was perceived moving in the distal direction (DI = pdistal − pproximal). DI could range from −1 to 1, with 1 indicating that perceived motion was always distal, −1 indicating that it was always proximal, and 0 indicating that the two directions were perceived equally often, suggesting a neutral percept. For the prolonged adaptation experiment the DI was calculated as the proportion of time the stimulus was perceived moving opposite to the adapting motion, subtracted from the proportion of time it was perceived moving in the same direction as the adapting motion.

RESULTS

Perceived direction of the test stimulus after adaptation.

In the 3-s adaptation experiment, the perception of the test stimulus was usually biased towards the direction opposite to the preceding adapting motion for each of the eight subjects (6 naïve and 2 authors who are referred to as P6 and P7 in the plots). Figure 4 shows the DI (see methods) of the test stimulus following each adapting stimulus for each participant. In most cases, adaptation to proximal motion resulted in a distal motion aftereffect [median DI = +0.58, interquartile range (IQR): 0.41, 0.83], but adaptation to distal motion did not always result in a proximal motion aftereffect (median DI = −0.17, IQR: −0.42, 0.05), reflecting an overall distal bias (although with apparent individual differences). The strength of the aftereffect (DI magnitude) was significantly greater for adaptation in the proximal direction (Wilcoxon signed rank test, n = 128, W+ = 1,667, P < 0.001). When the test stimulus did not have an immediately preceding adapting stimulus, the median DI was +0.26 (IQR: 0.07, 0.35), which was significantly greater than 0 (Wilcoxon signed rank test, n = 16, W+ = 113.5, P = 0.020), also displaying a distal bias.

Fig. 4.

Fig. 4.

Open in a new tab

Results from the 3-s adaptation experiment: perceived direction of the t-component test stimulus following adaptation, for each participant (P1–P8). The direction index (DI) is shown on the ordinate, with positive values indicating perceived direction was distal (D), negative values indicating perceived direction was proximal (P), and a value of 0 indicating a neutral percept. The adapting velocity is shown on the abscissa on a log scale, with negative values indicating adaptation in the proximal direction and positive values indicating adaptation in the distal direction. Adapting velocity = 0 indicates that there was no adapting stimulus immediately preceding the test stimulus. The 19 mm/s test speed data are shown as dashed lines with squares and 136 mm/s as solid lines with circles.

The bias to perceive distal motion was greater for the fast test stimulus than for the slow test stimulus. The fast test stimulus (136 mm/s) when there was no preceding adapting stimulus had a median DI of +0.34 (IQR: 0.00, 0.67), while the median DI for the slow test stimulus (19 mm/s) was +0.04 (IQR: −0.33, 0.09). This difference was statistically significant according to a Wilcoxon test with the DI scores for fast and slow test speeds paired for each adapting speed and participant (paired Wilcoxon signed rank test, median difference = −0.24, n = 72, W+ = 2,066.5, P < 0.001).

In the 10-s adaptation experiment, the distal bias was less evident, though still present (Fig. 5). Adaptation to both proximal and distal motion produced negative aftereffects: motion was perceived in the direction opposite that of the adapting stimulus. When adaptation was in the proximal direction, median DI was +0.78 (IQR: 0.50, 0.94), and when adaptation was in the distal direction, it was −0.57 (IQR: −0.78, −0.31). The difference in magnitude was significant, with proximal adaptation producing a stronger aftereffect (Wilcoxon signed rank test W+ = 2,484, n = 70, P < 0.001). Median DI for the test stimulus with no adaptation was +0.21 (IQR: −0.08, 0.45), which was significantly greater than 0 (Wilcoxon signed rank test, W+ = 297, n = 28, P = 0.03), also reflecting a distal bias. A trend was apparent for a greater distal bias for the fast (68 mm/s) test stimulus, but this was not statistically significant (Wilcoxon signed rank test, n = 84, W+ = 1,337, P = 0.092).

Fig. 5.

Fig. 5.

Open in a new tab

Results from the 10-s adaptation experiment: same format as Figure 4. The 34 mm/s test speed data are shown as dashed lines with squares and 65 mm/s as solid lines with circles.

The effect of adaptation speed.

To test for variation in the strength of the aftereffect due to adapting speed, the directional bias needed to be accounted for. To do this, the data from adaptation to both directions were combined into a DI difference score. That is, each participant's DI following proximal adaptation was subtracted from the DI following distal adaptation for each combination of adapting and test speeds. Any overall bias for proximal or distal responses should affect the DI for both proximal and distal adaptation equally, so subtracting the DIs should eliminate it. The DI difference score indicates the strength of the aftereffect: a score of 0 indicates no effect of the direction of the adapting stimulus on the perceived direction of the test stimulus. A negative value (minimum = −2) indicates an aftereffect in the expected direction (opposite to the adapting motion), and a positive value (maximum = 2) indicates an aftereffect in the unexpected direction (same as the adapting motion).

The median DI difference score for the test stimulus in the 3-s adaptation experiment was −0.70 (IQR: −1.00, −0.42), significantly less than 0 (Wilcoxon signed rank test, n = 64, W+ = 0, P < 0.001). This indicates that after accounting for the directional bias of each subject, there was a significant aftereffect in the predicted direction, opposite to that of the adapting stimulus. A similar result was observed for the 10-s adaptation experiment, with the median DI difference score equal to −1.25 (IQR: −1.63, −0.91), significantly less than 0 (Wilcoxon signed rank test, n = 70, W+ = 8, P < 0.001).

The relationship between DI difference scores and adapting speed can be seen in Fig. 6, which suggests that the strength of the tMAE simply increases with adapting speed, consistent with an intensive code for speed. To assess the effect of the test speed and the adapting speed, these factors were submitted to a repeated-measures ANOVA. The DI difference score is bounded by −2 and 2 and so violates the ANOVA assumption of a normal distribution; this was addressed by transforming the DI difference score by first rescaling it to span from 0 and 1 (adding 2 then dividing by 4) and then applying the logit transformation (inverse of the logistic) so that its possible range would span negative infinity to positive infinity. This transformed DI difference score was used as the dependent variable in the statistical analyses; however, untransformed DI difference scores are reported in the text and figures for ease of interpretation. Adapting speed had a significant effect (F3,21 = 8.0, P = 0.001, η2 = 0.52) according to the ANOVA, but test speed did not (F1,7 = 0.6, P = 0.454, η2 = 0.08), nor was there a significant interaction between adapting speed and test speed (F3,21 = 0.7, P = 0.544, η2 = 0.10).1

Fig. 6.

Fig. 6.

Open in a new tab

Results from the 3-s adaptation experiment: Perceived direction of the adapting stimuli and the test stimuli, averaged across participants. The abscissa shows the adapting speed on a log scale. The DI difference score given on the ordinate is an indicator of the strength and direction of the motion signal, with a value of 2 indicating the stimulus was always perceived in the direction of the adapting stimulus, and a value of −2 indicating the stimulus was always perceived in the direction opposite the adapting stimulus (see text). For the test stimulus, DI difference indexes tMAE strength, with more negative values indicating a stronger aftereffect. Left: The mean DI difference scores are plotted separately for the 2 test speeds, 19 mm/s (grey) and 136 mm/s (black). Right: data are averaged across the 2 test speeds, and the lines show the predicted values from the linear model (see text); they appear curved because a logit transformation was used for the regression. Error bars are 95% confidence intervals.

A planned linear contrast was conducted in the ANOVA with the contrast weights adjusted to reflect the nonequal spacing of the log of the adapting speeds used in the experiment. The linear contrast was significant (F1,7 = 24.5, P = 0.002, η2 = 0.78), indicating a positive relationship between adapting speed and the strength of the aftereffect (in effect, this assessed for the presence of an exponential rather than linear trend). This effect did not significantly interact with test speed (F1,7 = 0.6, P = 0.464, η2 = 0.08). As there was no difference for the two test speeds the transformed DI difference scores were combined for the two test speeds, and a regression line was fitted. The slope of this line as a function of the log adapting speed was −4.6 (95% confidence interval: −7.2, −2.0; see fitted line for the test stimulus in Fig. 6, right, “Test stimulus”).

The results of the prolonged adaptation experiment were consistent with the 3-s adaptation experiment with respect to the relationship between DI difference scores and the speed of the adapting and test stimuli. An ANOVA revealed that adapting speed had a significant effect on the transformed test DI difference score (F4,24 = 3.7, P = 0.017, η2 = 0.38), but test speed did not (F1,6 = 0.1, P = 0.736, η2 = 0.02), nor was there a significant interaction between adapting speed and test speed (F4,24 = 1.7, P = 0.187, η2 = 0.22). With the use of the same contrast analysis as in the 3-s adaptation experiment, the test DI significantly increased with adapting speed (F1,6 = 7.4, P = 0.035, η2 = 0.55), and this effect did not significantly interact with test speed (F1,6 = 2.7, P = 0.154, η2 = 0.31).

Perceived direction of the adapting stimuli.

The adapting stimuli had a single unambiguous direction (proximal or distal), so it was expected that the direction perceived would be at least as clear as it was for the aftereffects perceived with the two-component test stimuli, yielding a large magnitude for the DI difference for perception of the adapter. The median DI difference for the adapting stimuli in the 3-s adaptation experiment was +1.83 (IQR: +1.50, +1.92); the positive sign indicating that the direction perceived was veridical. This was significantly greater than the magnitude of the DI difference for the test stimulus (median test DI = −0.70, or a magnitude of 0.70; paired Wilcoxon signed rank test, n = 64, W+ = 2,080, P < 0.001), indicating perception of the adapting stimulus was more strongly directional than the test stimulus. Raw DI scores of the adapting stimuli indicate that unlike responses to the test stimuli (Fig. 4), responses to the adapting stimuli were at or close to ceiling performance (Fig. 7). Faster speeds resulted in more accurate direction judgments: the slope of the regression line fitted to the transformed DI difference scores for the adapting stimuli as a function of the log adapting speed was 2.4 (95% confidence interval: 0.6, 4.2; see fitted line for the adapting stimulus in Fig. 6, right, “Adapting stimulus”).

Fig. 7.

Fig. 7.

Open in a new tab

Results from the 3-s adaptation experiment: Individual data showing perceived direction of the adapting motion. The DI (see text for its calculation) is shown on the ordinate, with positive values indicating perceived direction was distal, negative values indicating perceived direction was proximal, and a value of 0 indicating a neutral percept. The speed of the adapting motion is shown on the abscissa on a log scale, with negative values indicating motion in the proximal direction and positive values indicating motion in the distal direction.

One possibility we wished to exclude is that the direction of the slow adapting stimuli was difficult to perceive, resulting in a weaker aftereffect than for the fast adapting stimuli. To evaluate this possibility we conducted a multiple regression analysis. Independent variables were adapting speed, DI for the adapting stimulus and test speed, and the dependent variable was DI for the test stimulus; “participant” was also included as an independent variable to account for the effect of individual differences. Each participant's mean test DI was used as his or her score across all conditions (a method for dealing with categorical variables in regression called '“criterion scaling”; see, e.g., Keith 2006, ch. 6). The participant variable was not statistically significant (F7,55 = 1.7, P = 0.121). Of the other three predictors, only adapting speed had a significant effect on DI for the test stimulus (R2 = 0.431, t59 = 4.17, P < 0.001). Importantly, DI for the adapting stimulus did not have a significant effect (R2 = 0.063, t59 = 0.59, P = 0.560). The test speed was also not significant (R2 = −0.157, t59 = −1.55, P = 0.126). The percentage of variance in the DI for the test stimulus explained by the adapting speed, DI for the adapting stimulus and test speed was 18.0, 2.4, and 2.5, respectively.

Prolonged adaptation.

The prolonged adaptation experiment involved adaptation using each combination of speeds of the adapting (23, 34, 45, 68, and 136 mm/s) and test (34 and 68 mm/s) stimuli. We tested for a general effect of prolonged adaptation, assessed as changes in the strength of the aftereffect over the repeated presentations of the test stimuli, by calculating the DI difference for each presentation. Linear regression revealed no significant trend over repeated presentations (slope = −0.01 ± 0.02, F1,347 = 0.7, P = 0.400).

Clarity of direction perception.

In the prolonged adaptation experiment, participants were given the option to report that the adapting and test stimuli were “unclear,” rather than only “distal” or “proximal” as in the 3-s adaptation experiment. For completeness, we report the proportions of the “unclear” response here. Participants reported the adapting stimulus as “unclear” relatively infrequently, with a mean proportion of 0.17 (range: 0.01–0.44). For the test stimulus, the proportion of unclear responses was similar for naive subjects (0.00–0.43) and authors (0.17, 0.18) when there was a preceding stimulus. However, there was an apparent difference between the authors and naïve participants in the judgments of the two-component test stimulus without any preceding directional adaptation. The two authors' proportions of “unclear” responses were 0.53 and 0.45, respectively, while for the five naïve participants, the proportion of “unclear” responses ranged from 0.00 to 0.31.

DISCUSSION

We used psychophysical judgments of direction of motion following a period of adaptation to evaluate two possibilities of how speed may be represented by direction-sensitive neurons in the human somatosensory system. Adapting stimuli were presented at a broad range of speeds, and the presence of the aftereffect (tMAE) was tested using four speeds in two experiments.

Our test stimuli were composed of interleaved lines of motion in two directions: one that matched the adapting stimulus and the other in the opposite direction but with the same speed. As reported previously (McIntyre et al. 2014), this combination of adapt-test stimulus reliably elicited a tMAE. Several of the participants appeared to have a bias to perceive the tMAE as in the distal direction, as evidenced by the relatively higher magnitude direction index when reporting tMAE after adapting to proximally directed motion than after adaptation to distally-directed motion (Fig. 3). Proximal-distal asymmetries in perception of tactile motion have been observed for stimuli applied to other body parts (Brugger and Meier 2015), but so far remain unexplained. In our study, the bias was dealt with by using the difference between the DI elicited from adapting stimuli in the proximal and distal directions to indicate adaptation strength irrespective of direction.

The results show that the tMAE is sensitive to the speed of the adapting stimulus, with greater speeds causing a stronger bias to perceive motion in the direction opposite to the adapting motion. However, we found no evidence that matching the adapting and the test speed affected the tMAE in either experiment. This is contrary to the predictions of the speed tuning theory, according to which matched adapt-test speeds should result in a stronger direction aftereffect than unmatched speeds. We expect that the range of speeds we used (from 19 to 136 mm/s) would have revealed speed tuning even if it was of a relatively broad bandwidth. Therefore our results support an intensive speed code model, where all neurons with a given direction selectivity show a similar change in their activity for changes in tactile speed.

A candidate speed cue that is coded intensively is the vibration energy created when textured surfaces move across the skin. A recent study suggests that it may be used for speed discrimination (Dallmann et al. 2015). However, the Optacon used in our study provides no such cue.

The pins of the Optacon used in the present study have a stereotyped protraction profile that reliably produces the same response from stimulated afferents on every cycle of the vibration (Gardner and Palmer 1989; Birznieks and Vickery 2013). The vibration energy of our stimulus is thus directly proportional to the number of pin protractions per second, and the frequency of the pin vibration was always constant at 230 Hz.

The apparent motion was generated by discrete steps of vibration along the skin with apparent speed determined by the number of cycles of vibration of each pin. Primary afferents adapt to vibration, and their adaptation level depends on frequency (Bensmaïa et al. 2005; Leung et al. 2005), but our protocol should not have resulted in differential adaptation for different adapting speeds or directions, because regardless of the stimulus speed, the array produced on average 10 pin protractions per second over the adaptation period.

However, while temporal frequency (TF) would typically be defined here as frequency of vibration when a pin was vibrating, we can also consider the time until a pin at a particular location was activated after its activation ceased. This “reactivation TF,” reflecting the time until the simulated moving object revisited its former location, was confounded with speed. Reactivation TF varied from 0.7 to 5 Hz for our speeds of 19 to 136 mm/s. Thus in principle, reactivation TF could have contributed to the adaptation, although it is less important than speed itself for determining responses of texture-independent motion-sensitive neurons in SI (see Fig. 4 in Dépeault et al. 2013), and also for perceptual speed judgments (see Fig. 7 in Dépeault et al. 2008). Moreover, we have recently shown that adapting speed and not reactivation TF determines the strength of the tactile speed aftereffect, in which the perceived speed of a moving surface is reduced following adaptation (McIntyre et al. 2016).

What neural adaptation sites might be responsible for the directional aftereffect we observed? Adaptation in primary afferents might have occurred but it would not be direction-selective and is thus unlikely to be the underlying mechanism of the tMAE, which is by definition direction dependent. Direction of motion in tactile stimuli lacking friction, such as the stimulation created by Optacon, is extracted by the central spatiotemporal integration of the input (Pei and Bensmaia 2014). It is thus likely that the adaptation underlying the tMAE occurred at tactile motion processing stages after the primary afferents, possibly in SI. This is supported by an functional MRI study, which found that experience of the tMAE is associated with activity in SI (Planetta and Servos 2012).

Because the strength of the tMAE increased with adapting speed, our study predicts that some neurons should be both direction-sensitive and more strongly driven at higher speeds. It appears these two characteristics do go together in SI: speed sensitive neurons are commonly reported to have direction selective responses (Whitsel et al. 1979; Essick and Whitsel 1985; Ruiz et al. 1995; Pei et al. 2010).

Although no neurophysiological study has estimated the proportion of neurons in SI with joint direction and speed coding, separate studies using recordings from single cortical neurons in monkeys indicate that ∼60% of SI motion-sensitive neurons are sensitive to the direction of motion (Warren et al. 1986) and that 64% of motion-sensitive SI neurons are sensitive to the speed of motion (Dépeault et al. 2013). While most of these speed-sensitive neurons, found in areas 3b, 1, and 2, also responded to texture (spatial period), a smaller number responded to speed irrespective of texture, and were found mostly in areas 1 and 2. All of these speed sensitive neurons increased their mean firing rate with speed over a similar range (40–100 mm/s) as used in our current study, with faster speeds producing a greater firing rate. Notably, direction selective neurons insensitive to texture are also found primarily in areas 1 and 2 (Pei et al. 2010).

Supporting the neural data, we found no perceptual evidence for speed tuning in direction-sensitive neurons. However, the existence of speed tuned neurons in tactile system cannot be ruled out, especially if there are stages where movement speed is processed independently from the directional information.

In summary, we show that a negative tMAE, elicited reliably by our two-component dynamic test stimulus, increases monotonically in strength with increases in adaptation speed over a range of speeds typically used during haptic exploration. This psychophysical finding is consistent with a lack of neurophysiological evidence from cortical areas 3b, 1 and 2 that speed is coded using tuned populations of neurons and instead lends weight to a substantial role for intensive coding of speed in direction selective neurons.

GRANTS

The study was also supported by the School of Psychology, University of Sydney, Australian Research Council ARC Discovery Project DP110104691 (to T. Seizova-Cajic and I. Birznieks), and Grants FT0990767 and DP140100952 (to A. O. Holcombe).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: S.M., I.B., R.M.V., A.O.H., and T.S.-C. conception and design of research; S.M. performed experiments; S.M. analyzed data; S.M., I.B., R.M.V., A.O.H., and T.S.-C. interpreted results of experiments; S.M. prepared figures; S.M. drafted manuscript; S.M., I.B., R.M.V., A.O.H., and T.S.-C. edited and revised manuscript; S.M., I.B., R.M.V., A.O.H., and T.S.-C. approved final version of manuscript.

ACKNOWLEDGMENTS

We thank Edward N. Crawford from School of Medical Sciences, University of New South Wales, Australia, for development of the Optacon interface.

Footnotes

1

When instead of calculating the DI difference score, we performed a repeated-measures ANOVA using transformed DI scores as the dependent variable, while including adapting direction as an additional factor in the analysis, and a similar pattern of results was found.

REFERENCES

  1. Barlow HB, Hill RM. Evidence for a physiological explanation of the waterfall phenomenon and figural after-effects. Nature 200: 1345–1347, 1963. [DOI] [PubMed] [Google Scholar]
  2. Bensmaïa SJ, Leung YY, Hsiao SS, Johnson KO. Vibratory adaptation of cutaneous mechanoreceptive afferents. J Neurophysiol 94: 3023–3036, 2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Birznieks I, Vickery RM. The role of temporal features of the afferent spike train in the perception of vibrotactile stimulus frequency. In: Neuroscience Meeting Planner. San Diego, CA: Soc Neurosci, 2013, p. 644.19/NN8. [Google Scholar]
  4. Bliss JC. Optical-to-tactile image conversion aids for the blind. Man-Machine Syst IEEE Trans 11: 58–65, 1970. [PubMed] [Google Scholar]
  5. Bradley DC, Goyal MS. Velocity computation in the primate visual system. Nat Rev Neurosci 9: 686–695, 2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Brugger P, Meier R. A new illusion at your elbow. Perception 44: 219–221, 2015. [DOI] [PubMed] [Google Scholar]
  7. Cavanagh P. Attention-based motion perception. Science 257: 1563, 1992. [DOI] [PubMed] [Google Scholar]
  8. Clifford CW. Perceptual adaptation: motion parallels orientation. Trends Cogn Sci 6: 136–143, 2002. [DOI] [PubMed] [Google Scholar]
  9. Clifford CW. The tilt illusion: phenomenology and functional implications. Vision Res 104: 3–11, 2014. [DOI] [PubMed] [Google Scholar]
  10. Dallmann CJ, Ernst MO, Moscatelli A. The role of vibration in tactile speed perception. J Neurophysiol 114: 3131–3139, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Dépeault A, Meftah EM, Chapman CE. Tactile speed scaling: contributions of time and space. J Neurophysiol 99: 1422–1434, 2008. [DOI] [PubMed] [Google Scholar]
  12. Dépeault A, Meftah EM, Chapman CE. Neuronal correlates of tactile speed in primary somatosensory cortex. J Neurophysiol 110: 1554–1566, 2013. [DOI] [PubMed] [Google Scholar]
  13. Essick GK, Bredehoeft KR, McLaughlin DF, Szaniszlo JA. Directional sensitivity along the upper limb in humans. Somatosens Mot Res 8: 13–22, 1991. [DOI] [PubMed] [Google Scholar]
  14. Essick GK, Rath EM, Kelly DG, James A, Murray RA, Bolanowski SJ. A novel approach for studying direction discrimiuation. In: Somesthesis and the Neurobiology of the Somatosensory Cortex, edited by Franźen O, Johansson RS, Terenius L. Basel, Switzerland: Birkhäuser Verlag, 1996, p. 59–72. [Google Scholar]
  15. Essick GK, Whitsel BL. Factors influencing cutaneous directional sensitivity: a correlative psychophysical and neurophysiological investigation. Brain Res Rev 10: 213–230, 1985. [DOI] [PubMed] [Google Scholar]
  16. Gardner EP, Palmer CI. Simulation of motion on the skin. I. Receptive fields and temporal frequency coding by cutaneous mechanoreceptors of OPTACON pulses delivered to the hand. J Neurophysiol 62: 1410–1436, 1989. [DOI] [PubMed] [Google Scholar]
  17. Gardner EP, Sklar BF. Discrimination of the direction of motion on the human hand: a psychophysical study of stimulation parameters. J Neurophysiol 71: 2414–2429, 1994. [DOI] [PubMed] [Google Scholar]
  18. Harris RA, O'Carroll DC, Laughlin SB. Contrast gain reduction in fly motion adaptation. Neuron 28: 595–606, 2000. [DOI] [PubMed] [Google Scholar]
  19. Hazlewood V. A note on failure to find a tactile motion aftereffect. Aust J Psychol 23: 59–62, 1971. [Google Scholar]
  20. Keith TZ. Multiple Regression and Beyond: An Introduction to Multiple Regression and Structural Equation Modeling. London: Routledge, 2014. [Google Scholar]
  21. Konkle T, Wang Q, Hayward V, Moore CI. Motion aftereffects transfer between touch and vision. Curr Biol 19: 745–750, 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kuroki S, Watanabe J, Mabuchi K, Tachi S, Nishida S. Directional remapping in tactile inter-finger apparent motion: a motion aftereffect study. Exp Brain Res 216: 311–320, 2012. [DOI] [PubMed] [Google Scholar]
  23. Lederman SJ. Tactile roughness of grooved surfaces: the touching process and effects of macro-and microsurface structure. Percept Psychophys 16: 385–395, 1974. [Google Scholar]
  24. Lederman SJ. Tactual roughness perception: spatial and temporal determinants. Can J Psychol 37: 498–511, 1983. [Google Scholar]
  25. Leung YY, Bensmaïa SJ, Hsiao SS, Johnson KO. Time-course of vibratory adaptation and recovery in cutaneous mechanoreceptive afferents. J Neurophysiol 94: 3037–3045, 2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lundström R, Johansson RS. Acute impairment of the sensitivity of skin mechanoreceptive units caused by vibration exposure of the hand. Ergonomics 29: 687–698, 1986. [DOI] [PubMed] [Google Scholar]
  27. Mather G, Moulden B. A simultaneous shift in apparent direction: further evidence for a “distribution-shift” model of direction coding. Q J Exp Psychol 32: 325–333, 1980. [DOI] [PubMed] [Google Scholar]
  28. McIntyre S, Seizova-Cajic T, Birznieks I, Holcombe A, Vickery R. Adaptation to motion presented with a tactile array. In: Haptics: Neuroscience, Devices, Modeling, and Applications SE-44, edited by Auvray M, Duriez C. Berlin, Germany: Springer, 2014, p. 351–359. [Google Scholar]
  29. McIntyre S, Seizova-Cajic T, Holcombe AO. The tactile speed aftereffect depends on the speed of adapting motion across the skin rather than other spatiotemporal features. J Neurophysiol 115: 1112–1121, 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Meftah EM, Belingard L, Chapman CE. Relative effects of the spatial and temporal characteristics of scanned surfaces on human perception of tactile roughness using passive touch. Exp Brain Res 132: 351–361, 2000. [DOI] [PubMed] [Google Scholar]
  31. Pack CC, Bensmaia SJ. Seeing and feeling motion: canonical computations in vision and touch. PLoS Biol 13: e1002271, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Pei YC, Bensmaia SJ. The neural basis of tactile motion perception. J Neurophysiol 112: 3023–3032, 2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Pei YC, Hsiao SS, Craig JC, Bensmaïa SJ. Shape invariant coding of motion direction in somatosensory cortex. PLoS Biol 8: e1000305, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Planetta PJ, Servos P. The tactile motion aftereffect revisited. Somatosens Mot Res 25: 93–99, 2008. [DOI] [PubMed] [Google Scholar]
  35. Planetta PJ, Servos P. The postcentral gyrus shows sustained fMRI activation during the tactile motion aftereffect. Exp Brain Res 216: 535–544, 2012. [DOI] [PubMed] [Google Scholar]
  36. Ruiz S, Crespo P, Romo R. Representation of moving tactile stimuli in the somatic sensory cortex of awake monkeys. J Neurophysiol 73: 525–537, 1995. [DOI] [PubMed] [Google Scholar]
  37. Schwartz O, Hsu A, Dayan P. Space and time in visual context. Nat Rev Neurosci 8: 522–535, 2007. [DOI] [PubMed] [Google Scholar]
  38. Smith AM, Gosselin G, Houde B. Deployment of fingertip forces in tactile exploration. Exp Brain Res 147: 209–218, 2002. [DOI] [PubMed] [Google Scholar]
  39. Thalman W. The after-effect of movement in the sense of touch. Am J Psychol 33: 268–276, 1922. [Google Scholar]
  40. Warren S, Hämäläinen HA, Gardner EP. Objective classification of motion- and direction- sensitive neurons in primary somatosensory cortex of awake monkeys. J Neurophysiol 56: 598–622, 1986. [DOI] [PubMed] [Google Scholar]
  41. Watanabe J, Hayashi S, Kajimoto H, Tachi S, Nishida S. Tactile motion aftereffects produced by appropriate presentation for mechanoreceptors. Exp Brain Res 180: 577–582, 2007. [DOI] [PubMed] [Google Scholar]
  42. Whitsel BL, Dreyer DA, Hollins M, Young MG. The coding of direction of tactile stimulus movement: Correlative psychophysical and electrophysiological data. In: Sensory Function of the Skin of Humans, Proceedings of the Second International Symposium on the Skin Senses, edited by Kenshalo DR. New York: Plenum Press, 1979, p. 79–107. [Google Scholar]
  43. Yoshioka T, Craig JC, Beck GC, Hsiao SS. Perceptual constancy of texture roughness in the tactile system. J Neurosci 31: 17603–17611, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Neurophysiology are provided here courtesy of American Physiological Society

RESOURCES