due is 24 hours

http://informahealthcare.com/smr ISSN: 0899-0220 (print), 1369-1651 (electronic)

Somatosens Mot Res, 2015; 32(3): 163–171 ! 2015 Informa UK Ltd. DOI: 10.3109/08990220.2015.1023950

O R I G I N A L A R T I C L E

Tactile processing in children and adolescents with obsessive–compulsive disorder

Burak Güçlü1, Canan Tanıdır2, Emre Çanayaz1,3, Bora Güner4, Hamiyet _Ipek Toz2, Özden Ş. Üneri2, & Mark Tommerdahl5

1Institute of Biomedical Engineering, Boğaziçi University, _Istanbul, Turkey, 2Child and Adolescent Psychiatry Clinic, Bakırköy Prof. Dr Mazhar Osman

Training and Research Hospital for Psychiatry, Neurology and Neurosurgery, _Istanbul, Turkey, 3Vocational School of Technical Sciences, Marmara

University, _Istanbul, Turkey, 4Department of Biomedical Engineering, Duke University, Durham, NC, USA, and 5Department of Biomedical

Engineering, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA

Abstract

Many obsessive–compulsive disorder (OCD) patients experience sensory phenomena, such as bodily sensations and ‘‘just-right’’ perceptions accompanying compulsions. We studied tactile processing in OCD by psychophysical experiments targeting the somatosensory cortex. Thirty- two children and adolescents with OCD (8 tic-related, 19 with sensory phenomena (SP)) and their sex- and age-matched controls participated in the study. After clinical assessments, two questionnaires were completed for sensory problems (Sensory Profile and Touch Inventory for Elementary-School-Aged Children). The psychophysical experiments consisted of five tasks: simple reaction time, choice reaction time, dynamic (detection) threshold, amplitude discrimination, and amplitude discrimination with single-site adaptation. The tactile stimuli were sinusoidal mechanical vibrations (frequency: 25 Hz) applied on the fingertips. Just- noticeable differences (JNDs) were found in amplitude discrimination tasks. There was no difference between the OCD group and controls in detection thresholds. However, the OCD group (especially young males) had worse amplitude discrimination (i.e., higher JNDs) than controls. Young OCD participants had reduced adaptation than young controls. Tic-related OCD participants and those with SP had higher detection thresholds than those without. Additionally, the OCD group reported more problems than controls in the Emotional/Social subset of the Sensory Profile questionnaire. The discrimination results show altered tactile processing in OCD at suprathreshold levels. This can be explained by a scaling factor modifying the sensory signal with decreasing slope at higher input levels to achieve normal Weber fractions internally. Quadratic discriminant analysis gave the best positive (76%) and negative (60%) predictive values for classifying individuals (into ‘‘OCD’’ or ‘‘control’’ groups) based on psychophysical data alone.

Keywords

Cortex, discriminant analysis, OCD, psychophysics, somatosensory, touch

History

Received 2 December 2014 Revised 16 February 2015 Accepted 23 February 2015 Published online 8 June 2015

Introduction

Childhood obsessive–compulsive disorder (OCD) affects

1–3% of the pediatric population (Thomsen 2013) and

interferes with their family, academic, and social activities

(Storch et al. 2010; Bipeta et al. 2013). In DSM-IV-TR

(Diagnostic and Statistical Manual of Mental Disorders,

American Psychiatric Association), OCD was listed under

anxiety disorders, whereas in DSM-5 it is classified separately

under a different category, named as ‘‘obsessive–compulsive

and related disorders’’. Additionally, a tic-related subtype was

introduced in DSM-5. Tic-related OCD patients were found to

have more touching, rubbing, blinking, and staring rituals, but

fewer cleaning rituals than non-tic-related OCD patients

(Holzer et al. 1994). A significant number of OCD patients

also report subjective experiences such as bodily sensations

(tactile, musculo-skeletal, or visceral), sense of inner tension,

feelings of incompleteness, and ‘‘just-right’’ auditory/visual/

tactile perceptions preceding or accompanying compulsions

(Prado et al. 2008). Recently there has been much clinical

interest on these experiences, now referred to as ‘‘sensory

phenomena’’ (SP). For example, Lee et al. (2009) found that

the occurrence of any kind of SP and its severity were higher

in OCD patients. At least one type of SP was reported by 65%

of the patients, and 16% of those described SP as more severe

than their obsessions (Ferrão et al. 2012).

Based on imaging studies and the main symptoms,

that is, obsessive doubts and repetitive behaviors, the neural

basis of OCD is currently considered to be some dysfunction

Correspondence: B. Güçlü, Institute of Biomedical Engineering, Boğaziçi University, Kandilli Campus, Çengelköy, _Istanbul 34684, Turkey. Tel: +90 216 5163467. Fax: +90 216 5163479. E-mail: [email protected]

in cortico-basal ganglia-thalamo-cortical pathways (Maia

et al. 2008). Imaging studies have shown reduced caudate

nucleus (Robinson et al. 1995; MacMaster et al. 2008) and

striatal (Rosenberg et al. 1997b) volume, increased gray

matter volume in anterior cingulate gyrus (Szeszko et al.

2004), and increased resting metabolic activity in orbito-

frontal cortex and basal ganglia (Swedo et al. 1992; Schwartz

et al. 1996). Functional studies also state the involvement

of these anatomical structures. Adult patients with OCD

had more response-suppression failures in oculomotor

tasks (Rosenberg et al. 1997a) and pediatric patients had

abnormal frontostriatal brain activation during tasks of

inhibition (Woolley et al. 2008). Acoustic prepulse inhibition

(Swerdlow et al. 1993; Hoenig et al. 2005; Ahmari et al.

2012) was reduced in adult patients, which implies reduced

inhibitory control. Similar results were obtained with tactile

prepuff inhibition in children with Tourette’s syndrome which

frequently has comorbidity with OCD (Swerdlow et al. 2001).

The converging evidence thus supports the ‘‘impaired

sensorimotor gating’’ hypothesis, which proposes that self-

repeating loops in the abovementioned structures mediate the

OCD symptoms, and this is paralleled by a failure of the

weaker prepulse to inhibit the startle response normally.

Although ‘‘impaired sensorimotor gating’’ as described

above does not necessarily imply impaired sensory gating,

Rossi et al. (2005) found that postcentral somatosensory-

evoked potentials were not gated due to movement in patients

with OCD. They also showed that movement-related sensory

gating was restricted to precentral potentials and was reduced

compared to controls. This was also supported by event-

related potentials linked to the prediction and suppression of

sensory input due to movement in an agency paradigm

(Gentsch et al. 2012). Therefore, depending on the particular

task the related sensory cortex may be indirectly affected as

well. It has been suggested that SP may help determine

clinical features of OCD and identify subtypes based on

pathological sensory processing (Miguel et al. 1997, 2000).

Somatosensory processing has a critical role in social,

communicative, and motor development and was found to

be impaired in neurodevelopmental disorders such as autism

(Cascio 2010). Studies showing cortical disinhibition and

those reporting SP symptoms in patients suggest a possible

link also between OCD and somatosensory processing.

The aim of this study was to test tactile processing in

children and adolescents with OCD by a psychophysical

experiment not involving sensorimotor or movement-related

sensory gating effects. The experiment used accurately

controlled mechanical vibrations applied on the fingertip.

Since this procedure mainly targets low-level sensory pro-

cessing, the results are important for understanding the

functional changes specifically in the somatosensory cortex.

We previously performed similar experiments with normal

and autistic children (Güçlü and Öztek 2007; Güçlü et al.

2007). Intensive laboratory testing of six male children with

autism did not show differences in the detection thresholds of

the Pacinian and Meissner mechanoreceptor systems. Later

testing with a portable stimulator device which allowed a

larger sample size showed reduced adaptation (Tannan et al.

2008) and reduced synchronized conditioning in temporal

order judgments (Tommerdahl et al. 2008). These results

implied, respectively, disinhibition and local underconnectiv-

ity in autism. For OCD, a general disinhibition in somato-

sensory cortex would be expected to produce lower detection

thresholds and relatively lower adaptation at suprathreshold

levels. Here we also attempted to classify participants by

discriminant analysis solely based on psychophysical data.

Due to the continuing interest in the search for possible

subtypes, which would help diagnosis, treatment, and under-

standing the pathophysiology of OCD (Leckman et al. 2010),

two sensory questionnaires were completed: Sensory Profile

(Dunn and Westman 1997) and Touch Inventory for

Elementary-School-Aged Children (Royeen and Fortune

1990). Additionally, we repeated some of the analyses by

dividing the participants into subgroups based on age, gender,

presence of tics, and tactile SP.

Materials and methods

Participants

Thirty-two (21 female, 11 male) children and adolescents

with OCD (mean age: 12.6 years, range: 7–18), seeking

treatment in the child and adolescent psychiatry clinic of a

hospital, specialized in neurological and psychiatric disorders,

participated in the study. Thirty-two sex- and age-matched

controls (mean age: 13.0 years, range: 8–18) were recruited

from local schools. The summary of the demographic data is

given in Table I. None of the participants in the control group

had a history of any psychiatric disorder. Although we did not

test for intelligence in this group, students with poor school

performance were excluded. OCD patients who had comorbid

psychotic disorder, mental retardation, pervasive develop-

mental disorder, and specific learning disorder were also not

included in the study. The experiments presented here do not

pose any harm and they adhere to the tenets of the Declaration

of Helsinki for testing human participants. The study was

approved by the Medical Ethics Committee of Bakırköy Prof.

Dr Mazhar Osman Training and Research Hospital for

Psychiatry, Neurology and Neurosurgery. Parents of all

participants signed informed consent forms prior to participa-

tion. Adolescents older than 12 years signed their own consent

forms as well. Additionally, the participants were divided into

subgroups based on age (7–12: younger vs. 13–18: older),

gender, presence of tics (8 OCD participants), and presence

of SP in the form of tactile obsessions/compulsions and

‘‘just-right’’ perceptions (19 OCD participants).

Clinical assessments

Clinical assessments were made by experienced child and

adolescent psychiatrists during interviews with both the

parents and participants. OCD diagnosis was according to

the DSM-IV-TR criteria. Participants on medication contin-

ued their regimens throughout the study. Psychiatric comor-

bidity was determined by using the Schedule for Affective

Disorders and Schizophrenia for School-Age Children—

Present and Lifetime Version—Turkish Version (K-SADS-

PL-T) (Gökler et al. 2004). The K-SADS-PL-T is a semi-

structured interview schedule designed to assess 32 psychi-

atric disorders in children and adolescents on the basis of

DSM-IV criteria. The Turkish version of the Wechsler

164 B. Güçlü et al. Somatosens Mot Res, 2015; 32(3): 163–171

Intelligence Scale for Children—Revised (WISC-R) was

given to test for intellectual abilities (Savaşır and Şahin

1995). All OCD participants had Full Scale IQ (FSIQ), Verbal

IQ (VIQ), and Performance IQ (PIQ) higher than 70.

Sensory questionnaires

We also gave out two questionnaires specifically developed

for testing sensory problems in children and adolescents. The

Sensory Profile (Dunn and Westman 1997) consists of 125

items grouped under the categories of Auditory, Visual,

Activity Level, Taste/Smell, Body Position, Movement,

Touch, and Emotional/Social. The questions were answered

by the parents (n¼15 for control, n¼16 for OCD in Table I). We scored the test based on the percentage of typical children

who displayed the given behavior. We presented the positive

scores, that is, the number of items with Likert-scale

responses greater than the behavioral percentage for each

individual item (Güçlü et al. 2007). Touch Inventory for

Elementary-School-Aged Children (Royeen and Fortune

1990) has 26 items for measuring tactile defensiveness and

was answered by the participants (n¼26 for control, n¼17 for OCD in Table I). We presented the raw scores; high (low)

score shows tactile hyper(hypo)-responsivity.

Tactile stimuli and psychophysical procedures

The tactile stimuli were sinusoidal mechanical vibrations

(frequency: 25 Hz) generated by a portable device (CM-4;

Cortical Metrics, Chapel Hill, NC, USA). The device has a

head unit on which each participant placed his/her hand with

four fingertips (of digits D2–D5) touching the plastic

contactor probes (diameter: 5 mm, static indentation:

0.5 mm) at adjustable locations (Holden et al. 2012).

Based on self-reports for handedness, we stimulated the

digits D2 and D3 of the non-dominant hand. The device was

controlled by custom software running on a laptop computer

and the participants responded by pressing mouse buttons

(with their dominant hands) according to the instructions of

each measurement protocol. Each experiment consisted of a

battery of five sequential protocols which overall lasted �1 h including short breaks. The protocols were similar to those

given in Zhang et al. (2011) and are summarized below. All

amplitude values were measured zero-to-peak.

(1) Simple reaction time (sRT). A single-cycle (duration:

40 ms) sinusoidal wave (amplitude: 300 mm) was applied to D2 in every trial after a random inter-trial interval (ITI:

2–7 s). The participant’s task was to press any mouse

button as soon as the stimulus was detected, and the

reaction time was recorded. Each test had 20 trials; the

average of five median reaction-time values was used in

the analyses. This protocol was not directly related with

tactile sensitivity, but served to get the participant

accustomed to the device.

(2) Choice reaction time (cRT). This protocol was for

training the participants on a two-alternative forced-

Table I. Demographic data and main results. a

Control OCD Statistics

Age 13.0 (2.9) 12.6 (3.0) t(62)¼0.55, p¼0.59 Sex (F/M) 21/11 21/11 Tic-related – 8 SP – 19 WISC-R

FSIQ – 96.4 (11.0) VIQ – 91.6 (12.7) PIQ – 101.0 (11.8)

Sensory profile b

(positive scores) Total 57.0 (17.5) 55.6 (25.6) t(27)¼0.18, p¼0.43 Touch 10.5 (4.5) 9.9 (5.3) t(29)¼0.37, p¼0.36 Emotional/social 12.1 (3.4) 15.1 (5.6) t(25)¼1.81, p¼0.04, d¼0.64

Touch inventory b

(raw scores) 41.0 (4.7) 43.9 (9.5) t(21)¼1.19, p¼0.12 Psychophysical test battery

c

sRT (ms) 390 (110) 431 (103) t(58)¼1.47, p¼0.07 cRT (ms) 670 (224) 675 (182) t(57)¼0.10, p¼0.46 RT_d (ms) 244 (132) 235 (134) t(56)¼0.25, p¼0.40 DT_c (mm) 10.2 (2.2) 10.1 (3.5) t(47)¼0.04, p¼0.49 AD (mm) 80.9 (35.5) 113.3 (53.1) t(44)¼2.69, p¼0.005, d¼0.72 cAD (mm) 152.6 (58.8) 167.2 (61.5) t(46)¼0.87, p¼0.20 AD_d (mm) 73.9 (58.8) 51.3 (76.2) t(34)¼1.11, p¼0.14

a The results are given as means and standard deviations in parentheses. Bold entries show statistical differences between participant groups. OCD: obsessive–compulsive disorder, SP: sensory phenomena, WISC-R: Wechsler Intelligence Scale for Children—Revised, FSIQ: Full Scale Intelligence Quotient, VIQ: Verbal Intelligence Quotient, PIQ: Performance Intelligence Quotient, sRT: simple reaction time, cRT: choice reaction time, RT_d: paired difference between simple and choice reaction times, DT_c: dynamic threshold corrected for choice reaction time, AD: just-noticeable difference in amplitude discrimination, cAD: just-noticeable difference in amplitude discrimination with single-site adaptation, AD_d: paired change in just-noticeable difference due to adaptation.

b Because of low response rate, the sample sizes were reduced: n¼15 for control and n¼16 for OCD in Sensory Profile, n¼26 for control and n¼17 for OCD in Touch Inventory. See text for the calculation of the scores.

c Missing data were due to non-convergent staircases, cue detection, or omission of outliers. The remaining sample sizes are as the following for the control and the OCD group, respectively: n¼30 and 30 in sRT, n¼31 and 30 in cRT, n¼29 and 29 in RT_d, n¼27 and 29 in DT_c, n¼31 and 27 in AD, n¼29 and 23 in cAD, n¼29 and 20 in AD_d.

DOI: 10.3109/08990220.2015.1023950 Tactile processing in children and adolescents with obsessive–compulsive disorder 165

choice (2AFC) task. It was similar to the previous

protocol except the stimulus was randomly either

presented to D2 or D3. The task was to press the correct

button (left button for D2, right for D3 in left-hand

stimulation) as soon as possible. The average measure-

ment was calculated as above, regardless of correct/

incorrect trials.

(3) Dynamic threshold (DT_c). This was a 2AFC in

which the participant detected the digit on which the

stimulus was randomly (D2 or D3) applied. After a

randomized delay period (0–3 s), the stimulus wave-

form was initiated, and the amplitude dynamically

increased from zero at a rate of 2 mm/s. The final amplitude setting at the instant of the button press was

recorded in each trial. There was a constant 5-s ITI

before the delay period started. Among seven trials

tested, the average of the values recorded in only the

correct trials was calculated as the dynamic threshold.

Next, this value was corrected by subtracting

the amplitude rise for the estimated reaction time in

this task.

(4) Amplitude discrimination (AD). Both digits D2 and D3

were stimulated with 25-Hz vibrations with durations

of 0.5 s. However, the standard stimulus always had the

amplitude of 200 mm, and the test stimulus had higher amplitude. The device randomly presented the test

stimulus to D2 or D3, and the participant’s task was to

discriminate and select the stronger stimulus by

pressing the associated button. Each trial ended with

a 5-s ITI. Initially, the test stimulus was easy to

discriminate (amplitude: 400 mm). Based on a staircase rule, the amplitude of the test stimulus was modified in

20-mm steps. In the first 10 trials, each correct response decreased the amplitude of the test stimulus

one step in the subsequent trial, and each incorrect

response increased it one step. For the next 10 trials,

two consecutive correct responses decreased it one

step, and one incorrect response increased it one step.

We only analyzed data which had convergent stair-

cases, that is, plus or minus one step fluctuation in the

last five trials. The just-noticeable difference (JND)

was calculated by subtracting the standard amplitude

from the average amplitude of the test stimulus in last

five trials.

(5) Amplitude discrimination with single-site adaptation

(cAD). This protocol was very similar to the previous

one except a 25-Hz adapting/conditioning stimulus

(amplitude: 200 mm, duration: 1 s) preceded the test stimulus. There was a 1-s empty interval between the

offset of the adapting stimulus and the onset of the test

stimulus. The effect of the adapting stimulus was to

mask the detection of the test, and therefore, increase

JND (without interfering with the standard stimulus).

There was, however, an important caveat. Some par-

ticipants realized that the adapting stimulus also acted

as a cue for predicting the presentation site of the test

stimulus, and quickly hit the lower limit of the staircase.

We did not use data from those participants and only

analyzed data which had convergent staircases as

explained above.

Data analysis

All statistical analyses were done in MS Excel 2007 and

MATLAB R2008a (MathWorks, Natick, MA, USA). The

outliers were eliminated by using Peirce’s criterion (Ross

2003). The remaining sample sizes for the questionnaire/

psychophysical data due to missing entries and outliers are

indicated in Table I. On this data set, we used one-tailed

t-tests for simple statistical comparisons between and within

participant groups/subgroups. Effect sizes were reported as

Cohen’s d. Next, we performed discriminant analyses on the

psychophysical data in order to classify the participants based

on tactile processing. To obtain a balanced multivariate data

set, we filled in the missing values of the psychophysical data

by gross averages (calculated by including both participant

groups). This made the classification problem more conser-

vative, that is, harder to discriminate between OCD partici-

pants and controls. Normality of the psychophysical variables

was tested by the Lilliefors test. Box’s M-test was used for

testing the homogeneity of the within-group covariance

matrices. Three classification methods were applied with

and without jackknife (i.e., leave-one-out cross-validation):

linear, quadratic, and allocation based on minimum

Mahalanobis distance computed by using distinct covariance

matrices (Manly 1986).

Results

Questionnaire data

The OCD participants were mostly in the normal range of the

IQs (Table I). Sensory Profile positive scores of OCD

participants and controls were similar. There was a statistical

difference between the participant groups in the Emotional/

Social subset of this questionnaire; OCD participants had

more problems on average (Table I). There was no statistical

difference between the participant groups in Touch Inventory

raw scores. Statistical analyses did not yield any significant

differences based on gender or age separately in control and

OCD groups, and based on the presence of tics or SP in the

OCD group, for the available questionnaire results reported

here. There were also no differences between the control and

OCD participants among either the female or male subgroups,

and among either the younger or older subgroups.

Psychophysical data

The means and standard deviations of the results obtained in

each psychophysical protocol are presented in Table I, as well

as the statistical comparisons between the participant groups

(see columns labeled ‘‘all’’ in Figures 1–3). In both groups,

the choice reaction time (cRT) protocol produced longer

reaction times compared to the simple reaction time (sRT)

protocol due to task difficulty (paired t(28)¼9.97, p50.001, d¼1.61 for control; paired t(28)¼9.46, p50.001, d¼1.73 for OCD). On average, there was a 244-ms increase for the

control and a 235-ms increase for the OCD group (RT_d).

However, there was no significant difference between the

participant groups in all listed reaction times. Both participant

groups almost had identical dynamic thresholds (DT_c:

�10 mm). The OCD group had much higher JND in the amplitude discrimination protocol compared to the control

166 B. Güçlü et al. Somatosens Mot Res, 2015; 32(3): 163–171

group (AD: 113.3 mm vs. 80.9 mm), which was also statistic- ally significant (Figure 3A). Both participant groups had

robust masking effects due to single-site adaptation. In other

words, the JNDs were higher with adaptation (cAD)

compared to those without (AD) (paired t(28)¼6.77, p50.001, d¼1.52 for control; paired t(19)¼3.01, p¼0.004, d¼0.89 for OCD). Specifically, the control and the OCD groups had average increases of 73.9 and 51.3 mm, respectively, in JNDs (AD_d). However, this change was not

statistically different between the participant groups (how-

ever, see below for younger participants). The participant

groups also had similar cAD values.

Participant subgroups

Data and statistical comparisons based on participant sub-

groups are presented in Figures 1–3. Repeating the above

statistical analyses did not yield any statistically significant

results between the older subgroups of the control and the

OCD participants. On the other hand, the younger control

subgroup had lower sRT value (444 ms) than that of the

younger OCD subgroup (515 ms) (Figure 1A). The difference

between AD values was more pronounced (control: 81.1 mm vs. OCD: 138.7 mm) for the younger subgroups (Figure 3A).

Moreover, the increase in JND after adaptation (AD_d) was

much higher in the younger control subgroup (84.9 mm) compared to the younger OCD subgroup (21.8 mm) (Figure 3C). There were also some age effects within each

Figure 1. Reaction-time results. (A) Simple reaction times (sRT), (B) choice reaction times (cRT), and (C) paired differences (RT_d) between sRT and cRT are plotted as averages of control (white columns) and OCD (black columns) participants. Error bars are the standard errors. ‘‘All’’ refers to all participants in the control and the OCD group. The remaining columns refer to the subgroups from each participant group. Every subgroup category based on age, gender, presence of tics, or presence of SPs is a partition of the entire participant group into two subgroups: young vs. old, female vs. male, no tic vs. (presence of) tic, or no SP vs. (presence of) SP. Significant differences are indicated as *p50.05, **p50.01, ***p50.001. For sRT: young control vs. young OCD (t(21)¼1.89, p¼0.04, d¼0.74), young vs. old control (t(23)¼2.48, p¼0.01, d¼0.93), young vs. old OCD (t(25)¼5.57, p50.001, d¼2.06), tic-related vs. non-tic-related (t(22)¼3.31, p¼0.002, d¼1.19). For cRT: young vs. old control (t(19)¼2.63, p¼0.008, d¼0.98), young vs. old OCD (t(21)¼4.04, p50.001, d¼1.52). For RT_d: female vs. male control (t(26)¼1.93, p¼0.03, d¼0.70).

Figure 2. Corrected dynamic thresholds (DT_c) are plotted as averages of control (white columns) and OCD (black columns) participants. See the caption of Figure 1 for additional information about the graph. No tic vs. tic (t(7)¼2.31, p¼0.03, d¼1.14), no SP vs. SP (t(24)¼2.20, p¼0.02, d¼0.77).

DOI: 10.3109/08990220.2015.1023950 Tactile processing in children and adolescents with obsessive–compulsive disorder 167

participant group. Among the control participants, the

younger subgroup had longer reaction time than the older

subgroup. Similar results were obtained for the older and

younger OCD subgroups as well (Figure 1A and B).

Additionally, among the OCD participants, the older sub-

group had lower JND in the AD task than the younger

subgroup (93.1 mm vs. 138.7 mm) (Figure 3A). Among same-gender subgroups, the only significant dif-

ference between control and OCD participants was in the AD

task. OCD males had higher JND compared to control males

(127.6 mm vs. 70.5 mm). We also analyzed the data based on gender separately for the control and the OCD groups, and

based on the presence of tics and SP in the OCD group. In the

control group, gender effects were found in RT_d and cAD

values. Female control participants had a higher increase from

sRT to cRT (273 ms) compared to males (189 ms). Similarly,

females (168.7 mm) had higher JND than males (122.0 mm) in amplitude discrimination with adaptation. This latter effect

also existed in the OCD participants. Female OCD partici-

pants had worse discrimination, that is, higher cAD value than

males (185.0 mm vs. 139.6 mm). In the OCD group, the presence of tics changed sRT and DT_c values. Tic-related

OCD participants had higher sRT value than the rest of the

OCD group (504 ms vs. 404 ms), and tic-related OCD

participants had higher dynamic threshold (13.3 mm vs. 9.1 mm). Dynamic threshold was also higher with SP in OCD participants compared to without SP (11.0 mm vs. 8.7 mm).

Discriminant analyses

Discriminant analyses were performed to study how well the

participants could be allocated to the correct group (control

vs. OCD) based solely on psychophysical data (Table II).

First, we tested the normality of each psychophysical variable

individually for participant groups (Lilliefors test). In the

control group, sRT, cRT, RT_d, and DT_c violated normality.

In the OCD group, DT_c, AD, cAD, and AD_d were not

distributed normally either. Since univariate normality could

not be established, it was not necessary to test for multivariate

normality. By using Box’s M-test, we found that the within-

group covariance matrices were statistically different

(M¼104, F(28, 13 400)¼3.27, p50.001), partially because this test is sensitive to normality violation. Since most of the

significance tests strictly assume multivariate normality and

homogeneity of covariance matrices, we cannot report the

statistical significance of the discriminant analyses. However,

the performances of the classification methods can be

Figure 3. Amplitude discrimination results. (A) Just-noticeable differences (JNDs) in amplitude discrimination (AD), (B) JNDs in amplitude discrimination with single-site adaptation (cAD), and (C) paired JND differences (AD_d) between AD and cAD are plotted as averages of control (white columns) and OCD (black columns) participants. See caption of Figure 1 for additional information about the graph. For AD: control vs. OCD (t(44)¼2.69, p¼0.005, d¼0.72), young control vs. young OCD (t(20)¼2.96, p¼0.004, d¼1.18), young vs. old OCD (t(20)¼2.35, p¼0.01, d¼0.92), male control vs. male OCD (t(15)¼2.43, p¼0.01, d¼1.07). For cAD: female vs. male control (t(25)¼2.45, p¼0.01, d¼0.90), female vs. male OCD (t(21)¼1.96, p¼0.03, d¼0.81). For AD_d: young control vs. young OCD (t(15)¼2.22, p¼0.02, d¼0.99).

168 B. Güçlü et al. Somatosens Mot Res, 2015; 32(3): 163–171

compared. Sensitivity refers to the percentage of correctly

allocated OCD participants in the OCD group. Specificity

refers to the percentage of correctly allocated controls in the

control group.

In the analyses without jackknife, each participant was

used in the classification model (i.e., training set); therefore, it

was slightly more likely for that participant to be allocated to

the correct group. With this method, the highest sensitivity

was obtained by classification based on minimum

Mahalanobis distances (81%). The highest specificity was

obtained by quadratic discriminant analysis (97%). The

analyses were repeated with jackknife classification, that is,

leave-one-out cross-validation, in which the participant being

allocated was not used for setting up the classification model.

Again, the highest sensitivity was obtained by minimum

Mahalanobis distances (62%), and the highest specificity was

obtained by quadratic discriminant analysis (88%).

Discussion

According to our knowledge, this study is the first to

psychophysically investigate tactile processing in OCD. For

stimuli at suprathreshold levels, our results are compatible

with the idea of cortical disinhibition in OCD which caused

lower adaptation in the amplitude discrimination task with

single-site adaptation. This was statistically significant in the

younger subgroup. Although the tactile sensitivity, as

measured by dynamic detection thresholds, was similar in

both OCD group and controls, OCD participants had worse

amplitude discrimination than controls. A stronger explan-

ation for these three main findings (similar detection thresh-

old, worse discrimination, reduced adaptation) is a scaling

factor which modifies the incoming sensory signal in OCD. In

order to achieve normal Weber fractions internally (Güçlü

2007), this factor should change nonlinearly, with a decreas-

ing slope at higher input levels. This preliminary model needs

much more effort for verification, but one can hypothesize on

some interesting predictions. For example, the scaling factor

is expected to be close to unity at low-level inputs, close to the

detection threshold. If this assumption is correct, the scaling

factor is likely to decrease from unity at higher inputs. It is

important to note that impaired sensorimotor gating measured

by the prepulse inhibition paradigm is probably not

influenced by a dysfunction described as above, because it

can be observed at different modalities (Swerdlow et al. 1993,

2001; Hoenig et al. 2005; Ahmari et al. 2012). Moreover, the

startle response without the prepulse is similar in OCD

participants and controls (Swerdlow et al. 1993). It would be

interesting to test intensity scaling in future studies (e.g., see

Güçlü and Dinçer 2013) for quantifying the range of the

hypothesized scaling factor in OCD.

Because of the high OCD comorbidity, patients diagnosed

with Tourette’s syndrome may also offer additional ideas

about the pathophysiology of OCD (Cohen et al. 2013). As a

matter of fact, similar results have been reported in the

literature regarding sensorimotor gating in Tourette’s syn-

drome (Smith and Lees 1989; Swerdlow et al. 1993, 2001;

Castellanos et al. 1996; Ziemann et al. 1997; Greenberg et al.

2000). Premonitory urges associated with motor and vocal

tics in Tourette’s syndrome are considered as SP, and they are

reported to be more bothersome than tics. Some Tourette’s

syndrome patients also feel heightened sensitivity to tactile,

auditory, and/or visual stimuli (Cohen and Leckman 1992;

Belluscio et al. 2011). Belluscio et al. (2011) subsequently

used Semmes–Weinstein monofilaments to test tactile sensi-

tivity and intensity scaling. They found no differences in

detection thresholds at lower leg and at tic sites. Higher

thresholds observed in our study, with tic-related OCD

participants and those with SP are somewhat contradictory

to Belluscio et al.’s (2011) results, if we assume a similar

pathophysiology in OCD with tics/SP and Tourette’s syn-

drome. We think this discrepancy may be due to the

inadequacy of mechanical stimulus control in their study.

The significant differences we found are numerically small

(13.3 mm for the tic-related OCD subgroup vs. 9.1 mm for the rest of the OCD participants; 11.0 mm for the OCD subgroup with SP vs. 8.7 mm for the OCD subgroup without SP). Semmes–Weinstein monofilaments cannot produce such

small differences, but these may be relevant for understanding

tactile processing (e.g., see Gescheider et al. 2005; Güçlü

et al. 2005). Belluscio et al. (2011) also observed that

Tourette’s syndrome patients used a lower range of numbers

compared to controls for intensity rating of near-threshold

tactile stimuli. This finding may be considered as consistent

with the decreasing scaling factor hypothesized above.

There is some evidence in our data that tactile processing

is disrupted in young male children with OCD more than any

other subgroup, both in terms of increased JND and reduced

adaptation in amplitude discrimination. Therefore, it would be

more efficient to extend the current study with homogeneous

groups. Due to ethical and clinical reasons, we could not

control the medication in OCD participants. Experimenting

with drug naive patients would ensure a relatively more

homogeneous group, and also verify that the significant

effects found in this study are not influenced by medication.

The abundance of those significant effects in subgroups based

on age, gender, presence of tics, and SP strengthens the value

of clinical and computational studies targeting subtypes in the

broad phenotypic variation of OCD (McElroy et al. 1994;

Stewart et al. 2007).

Both OCD and control participants who are older

responded faster to tactile stimuli than their younger

comparisons. This does not pose a problem regarding the

interpretation of the main results, because JNDs are inde-

pendent of reaction time. The dynamic threshold measure-

ments, which are slightly confounded by the slow rise in the

stimulus amplitude, were corrected for all participants’ own

Table II. Discriminant analyses.

Sensitivity Specificity PPV NPV

Without jackknife Linear 56 66 62 60 Quadratic 50 97 94 66 Mahalanobis 81 59 67 76

With jackknife Linear 56 59 58 58 Quadratic 41 88 76 60 Mahalanobis 62 31 48 45

Values are percentages. PPV: positive predictive value, NPV: negative predictive value.

DOI: 10.3109/08990220.2015.1023950 Tactile processing in children and adolescents with obsessive–compulsive disorder 169

reaction times. No significant effects were found in DT_c

values except in the presence of tics and SP. The average

reaction time of tic-related OCD participants is somewhat

higher (�100 ms) than non-tic-related OCD participants, but this would only cause a 0.2-mm increase in the threshold which cannot account for the large difference in the data

(4.2 mm). Therefore, threshold measurements are quite reli- able. The significant differences in reaction-time measure-

ments may be related to executive function and motor control,

which are out of our current scope.

The only significant difference between the participant

groups in the questionnaire results was the Emotional/Social

subset of the Sensory Profile. We performed preliminary

correlation analyses. OCD participants had more significant

and positive correlations between the Sensory Profile subset

items, but Touch Inventory scores of both groups were not

related to the Sensory Profile scores. Interestingly, dynamic

thresholds of the OCD group were highly correlated with the

Emotional/Social subset of Sensory Profile (r¼0.69, p¼0.006). We previously reported a similar link between tactile processing and emotions in autism, but only within the

questionnaire data (Güçlü et al. 2007). The current evidence

is more compelling and deserves a separate study utilizing

emotional modulation. It may be argued that the scarcity of

significant differences in the questionnaire results can be

confounded by the under-reporting of internal experiences by

young children (Banaschewski et al. 2003). However, our

analyses on the younger and the older subgroups showed that

this factor did not influence our results.

It seems that the classification method based on

Mahalanobis distances is the best for correctly identifying

OCD participants (81% sensitivity without jackknife) and the

quadratic method is the best for identifying controls (97%

specificity without jackknife). However, without any prior

knowledge the quadratic method gives the best positive and

negative predictive values (PPV and NPV) according to the

classification model presented here (prevalence: 50%). For an

undiagnosed individual classified into the ‘‘OCD’’ group

purely based on previous psychophysical data, there is 76%

probability that he/she has OCD. On the other hand, someone

who is placed into the ‘‘control’’ group as such has 60%

probability of being without OCD. One would expect a better

classification model if the training set included more partici-

pants. The calculations can be revised for the actual

prevalence to find PPV and NPV, but the purpose of the

discriminant analyses presented here was not to help in

diagnosis. The relatively good classification results are the

indication of the overall multivariate difference between OCD

and control participants. Since most of the pairwise compari-

sons between the participant groups did not yield significant

differences, this multivariate view is interesting.

The psychophysical protocols were optimized for speed, so

that a participant could be tested in a short session. The

biggest benefit of the stimulator device was portability; it

could be used in the clinical setting. Some of the protocols

may easily be revised for conforming to the traditional

research practice in psychophysics. For example, we plan to

measure detection thresholds by a 2AFC task with a staircase

procedure similar to that used for AD in the future. The real

challenge would be to selectively activate different receptor

systems (Güçlü and Bolanowski 2005) for understanding their

relative contributions to impaired amplitude discrimination

and reduced adaptation. This line of research may offer new

insights into the pathophysiology and treatment of OCD.

Acknowledgements

This study was supported by Boğaziçi University Research

Fund (BAP) project no. 13XP8 to Burak Güçlü. The authors

would also like to thank Albert A. Salah (Boğaziçi

University) for additional support (BAP project no. 6747S)

and helpful discussions.

Declaration of interest

Mark Tommerdahl is a co-founder of Cortical Metrics, LLC,

which designed and fabricated the CM-4 stimulator used in

this study. The rest of the authors report no conflicts of

interest. The authors alone are responsible for the content and

writing of the paper.

References

Ahmari SE, Risbrough VB, Geyer MA, Simpson HB. 2012. Impaired sensorimotor gating in unmedicated adults with obsessive–compulsive disorder. Neuropsychopharmacology 37:1216–1223.

Banaschewski T, Woerner W, Rothenberger A. 2003. Premonitory sensory phenomena and suppressibility of tics in Tourette syndrome: Developmental aspects in children and adolescents. Dev Med Child Neurol 45:700–703.

Belluscio BA, Jin L, Watters V, Lee TH, Hallett M. 2011. Sensory sensitivity to external stimuli in Tourette syndrome patients. Mov Disord 26:2538–2543.

Bipeta R, Yerramilli SS, Pingali S, Karredla AR, Ali MO. 2013. A cross- sectional study of insight and family accommodation in pediatric obsessive–compulsive disorder. Child Adolesc Psychiatry Ment Health 7:20. doi:10.1186/1753-2000-7-20.

Cascio CJ. 2010. Somatosensory processing in neurodevelopmental disorders. J Neurodev Disord 2:62–69.

Castellanos FX, Fine EJ, Kaysen D, Marsh WL, Rapoport JL, Hallett M. 1996. Sensorimotor gating in boys with Tourette’s syndrome and ADHD: Preliminary results. Biol Psychiatry 39:33–41.

Cohen AJ, Leckman JF. 1992. Sensory phenomena associated with Gilles de la Tourette’s syndrome. J Clin Psychiatry 53:319–323.

Cohen SC, Leckman JF, Bloch MH. 2013. Clinical assessment of Tourette syndrome and tic disorders. Neurosci Biobehav Rev 37: 997–1007.

Dunn W, Westman K. 1997. The Sensory Profile: The performance of a national sample of children without disabilities. Am J Occup Ther 51: 25–34.

Ferrão YA, Shavitt RG, Prado H, Fontenelle LF, Malavazzi DM, de Mathis MA, Hounie AG, Miguel EC, do Rosário MC. 2012. Sensory phenomena associated with repetitive behaviors in obsessive–com- pulsive disorder: An exploratory study of 1001 patients. Psychiatry Res 197:253–258.

Gentsch A, Schütz-Bosbach S, Endrass T, Kathmann N. 2012. Dysfunctional forward model mechanisms and aberrant sense of agency in obsessive–compulsive disorder. Biol Psychiatry 71: 652–659.

Gescheider GA, Güçlü B, Sexton JL, Karalunas S, Fontana A. 2005. Spatial summation in the tactile sensory system: Probability summa- tion and neural integration. Somatosens Mot Res 22:255–268.

Gökler B, Ünal F, Pehlivantürk B, Kültür EÇ, Akdemir D, Taner Y. 2004. Reliability and validity of Schedule for Affective Disorders and Schizophrenia for School Age Children—Present and Lifetime Version—Turkish Version (K-SADS-PL-T). Turk J Child Adolesc Ment Health 11:109–116 (in Turkish).

Greenberg BD, Ziemann U, Corá-Locatelli G, Harmon A, Murphy DL, Keel JC, Wassermann EM. 2000. Altered cortical excitability in obsessive–compulsive disorder. Neurology 54:142–147.

170 B. Güçlü et al. Somatosens Mot Res, 2015; 32(3): 163–171

Güçlü B, Bolanowski SJ. 2005. Vibrotactile thresholds of the Non- Pacinian I channel: I. Methodological issues. Somatosens Mot Res 22: 49–56.

Güçlü B, Gescheider GA, Bolanowski SJ, _Istefanopulos Y. 2005. Population-response model for vibrotactile spatial summation. Somatosens Mot Res 22:239–253.

Güçlü B. 2007. Deviation from Weber’s law in the Non-Pacinian I tactile channel: A psychophysical and simulation study of intensity discrim- ination. Neural Comput 19:2638–2664.

Güçlü B, Öztek Ç. 2007. Tactile sensitivity of children: Effects of frequency, masking, and the Non-Pacinian I psychophysical channel. J Exp Child Psychol 98:113–130.

Güçlü B, Tanidir C, Mukaddes NM, Ünal F. 2007. Tactile sensitivity of normal and autistic children. Somatosens Mot Res 24:21–33.

Güçlü B, Dinçer ŞM. 2013. Neural coding in the Non-Pacinian I tactile channel: A psychophysical and simulation study of magnitude estimation. Somatosens Mot Res 30:1–15.

Hoenig K, Hochrein A, Quednow BB, Maier W, Wagner M. 2005. Impaired prepulse inhibition of acoustic startle in obsessive–compul- sive disorder. Biol Psychiatry 57:1153–1158.

Holden JK, Nguyen RH, Francisco EM, Zhang Z, Dennis RG, Tommerdahl M. 2012. A novel device for the study of somatosensory information processing. J Neurosci Methods 204:215–220.

Holzer JC, Goodman WK, McDougle CJ, Baer L, Boyarsky BK, Leckman JF, Price LH. 1994. Obsessive–compulsive disorder with and without a chronic tic disorder. A comparison of symptoms in 70 patients. Br J Psychiatry 164:469–473.

Leckman JF, Denys D, Simpson HB, Mataix-Cols D, Hollander E, Saxena S, Miguel EC, Rauch SL, Goodman WK, Phillips KA, et al. 2010. Obsessive–compulsive disorder: A review of the diagnostic criteria and possible subtypes and dimensional specifiers for DSM-V. Depress Anxiety 27:507–527.

Lee JC, Prado HS, Diniz JB, Borcato S, da Silva CB, Hournie AG, Miguel EC, Leckman JF, do Rosário MC. 2009. Perfectionism and sensory phenomena: Phenotypic components of obsessive–compulsive disorder. Compr Psychiatry 50:431–436.

MacMaster FP, O’Neill J, Rosenberg DR. 2008. Brain imaging in pediatric obsessive compulsive disorder. J Am Acad Child Adolesc Psychiatry 47:1262–1272.

Maia TV, Cooney RE, Peterson BS. 2008. The neural bases of obsessive– compulsive disorder in children and adults. Dev Psychopathol 20: 1251–1283.

Manly BFJ. 1986. Multivariate statistical methods: A primer. London: Chapman & Hall.

McElroy SL, Phillips KA, Keck Jr PE. 1994. Obsessive compulsive spectrum disorder. J Clin Psychiatry 55:S33–S51.

Miguel EC, Baer L, Coffey BJ, Rauch SL, Savage CR, O’Sullivan RL, Phillips K, Moretti C, Leckman JF, Jenike MA. 1997. Phenomenological differences appearing with repetitive behaviours in obsessive–compulsive disorder and Gilles de la Tourette’s syndrome. Br J Psychiatry 170:140–145.

Miguel EC, do Rosário-Compos MS, Prado HS, do Valle R, Rauch SL, Coffey BJ, Baer L, Savage CR, O’Sullivan RL, Jenike MA, et al. 2000. Sensory phenomena in obsessive–compulsive disorder and Tourette’s disorder. J Clin Psychiatry 61:150–156.

Prado HS, Rosário MC, Lee J, Hounie AG, Shavitt RG, Miguel EC. 2008. Sensory phenomena in obsessive–compulsive disorder and tic disorders: A review of the literature. CNS Spectr 13:425–432.

Robinson D, Wu H, Munne RA, Ashtari M, Alvir JM, Lerner G, Koreen A, Cole K, Bogerts B. 1995. Reduced caudate nucleus volume in obsessive–compulsive disorder. Arch Gen Psychiatry 52:393–398.

Rosenberg DR, Dick EL, O’Hearn KM, Sweeney JA. 1997a. Response- inhibition deficits in obsessive–compulsive disorder: An indicator of dysfunction in frontostriatal circuits. J Psychiatry Neurosci 22:29–38.

Rosenberg DR, Keshavan MS, O’Hearn KM, Dick EL, Bagwell WW, Seymour AB, Montrose DM, Pierri JN, Birmaher B. 1997b. Frontostriatal measurement in treatment-naı̈ve children with obses- sive–compulsive disorder. Arch Gen Psychiatry 54:824–830.

Ross SM. 2003. Peirce’s criterion for the elimination of suspect experimental data. J Eng Technol 20:38–41.

Rossi S, Bartalini S, Ulivelli M, Mantovani A, Di Muro A, Goracci A, Castrogiovanni P, Battistini N, Passero S. 2005. Hypofunctioning of sensory gating mechanisms in patients with obsessive–compulsive disorder. Biol Psychiatry 57:16–20.

Royeen CB, Fortune JC. 1990. Touch inventory for elementary-school- aged children. Am J Occup Ther 44:155–159.

Savaşır I, Şahin N. 1995. Wechsler Çocuklar için Zeka Ölçeği (WISC-R) Uygulama Kitapçığı. Ankara: Turkish Psychological Association (in Turkish).

Schwartz JM, Stoessel PW, Baxter Jr LR, Martin KM, Phelps ME. 1996. Systematic changes in cerebral glucose metabolic rate after successful behavior modification treatment of obsessive–compulsive disorder. Arch Gen Psychiatry 53:109–113.

Smith SJM, Lees AJ. 1989. Abnormalities of the blink reflex in Gilles de la Tourette syndrome. J Neurol Neurosurg Psychiatry 52:895–898.

Stewart SE, Rosario MC, Brown TA, Carter AS, Leckman JF, Sukhodolsky D, Katsovitch L, King R, Geller D, Pauls DL. 2007. Principal components analysis of obsessive–compulsive disorder symptoms in children and adolescents. Biol Psychiatry 61:285–291.

Storch EA, Larson MJ, Muroff J, Caporino N, Geller D, Reid JM, Morgan J, Jordan P, Murphy TK. 2010. Predictors of functional impairment in pediatric obsessive–compulsive disorder. J Anxiety Disord 24:275–283.

Swedo SE, Pietrini P, Leonard HL, Shapiro MB, Rettew DC, Goldberger EL, Rapoport SI, Rapoport JL, Grady CL. 1992. Cerebral glucose metabolism in childhood-onset obsessive–compulsive disorder. Revisualizing during pharmacotherapy. Arch Gen Psychiatry 49: 690–694.

Swerdlow NR, Benbow CH, Zisook S, Geyer MA, Braff DL. 1993. A preliminary assessment of sensorimotor gating in patients with obsessive compulsive disorder. Biol Psychiatry 33:298–301.

Swerdlow NR, Karban B, Ploum Y, Sharp R, Geyer MA, Eastvold A. 2001. Tactile prepuff inhibition of startle in children with Tourette’s syndrome: In search of an ‘‘fMRI-friendly’’ startle paradigm. Biol Psychiatry 50:578–585.

Szeszko PR, MacMillan S, McMeniman M, Chen S, Baribault K, Lim KO, Ivey J, Rose M, Banerjee SP, Bhandari R, et al. 2004. Brain structural abnormalities in psychotropic drug-naı̈ve pediatric patients with obsessive–compulsive disorder. Am J Psychiatry 161:1049–1056.

Tannan V, Holden JK, Zhang Z, Baranek GT, Tommerdahl MA. 2008. Perceptual metrics of individuals with autism provide evidence for disinhibition. Autism Res 1:223–230.

Thomsen PH. 2013. Obsessive–compulsive disorders. Eur Child Adolesc Psychiatry 22:S23–S28.

Tommerdahl M, Tannan V, Holden JK, Baranek GT. 2008. Absence of stimulus-driven synchronization effects on sensory perception in autism: Evidence for local underconnectivity? Behav Brain Funct 4: 19. doi:10.1186/1744-9081-4-19.

Woolley J, Heyman I, Brammer M, Frampton I, McGuire PK, Rubia K. 2008. Brain activation in paediatric obsessive compulsive disorder during tasks of inhibitory control. Br J Psychiatry 192:25–31.

Zhang Z, Francisco EM, Holden JK, Dennis RG, Tommerdahl M. 2011. Somatosensory information processing in the aging population. Front Aging Neurosci 3:18. doi:10.3389/fnagi.2011.00018.

Ziemann U, Paulus W, Rothenberger A. 1997. Decreased motor inhibition in Tourette’s disorder: Evidence from transcranial magnetic stimulation. Am J Psychiatry 154:1277–1284.

DOI: 10.3109/08990220.2015.1023950 Tactile processing in children and adolescents with obsessive–compulsive disorder 171

Copyright of Somatosensory & Motor Research is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.