Somatosensory and Motor Research, March–June 2007; 24(1–2): 21–33 Tactile sensitivity of normal and autistic children BURAK GU ¨ C ¸ LU ¨ 1 , CANAN TANIDIR 2 , NAHIT MOTAVALLI MUKADDES 2 ,& FATIH U ¨ NAL 3 1 Biomedical Engineering Institute, Bog ˘azic ¸i University, Istanbul, Turkey, 2 Child and Adolescent Psychiatry Department, Autism Clinic, Istanbul Medical School, Istanbul University, Istanbul, Turkey, and 3 Child and Adolescent Psychiatry Department, Hacettepe University, Ankara, Turkey (Received 4 June 2006; revised 20 October 2006; accepted 11 December 2006) Abstract Many children with autistic spectrum disorders have unusual reactions to certain sensory stimuli. These reactions vary along a hyper- to hypo-responsivity continuum. For example, some children overreact to weak sensory input, but others do not respond negatively to even strong stimuli. It is typically assumed that this deviant responsivity is linked to sensitivity, although the particular stage of sensory processing affected is not known. Psychophysical vibrotactile thresholds of six male children (age: 8–12) who were diagnosed to have autistic spectrum disorders and six normal male children (age: 7–11) were measured by using a two-alternative forced-choice task. The tactile stimuli were sinusoidal displacements and they were applied on the terminal phalanx of the left middle finger of each subject. By using a forward-masking paradigm, 40- and 250-Hz thresholds of the Pacinian tactile channel and 40-Hz threshold of the Non-Pacinian I tactile channel were determined. There was no significant difference between the thresholds of autistic and normal children, and the autistic children had the same detection and masking mechanisms as the normal children. The sensory responsivity of each subject was tested by clinical questionnaires, which showed again no difference between the two subject groups. Furthermore, no significant correlations could be found between the questionnaire data and the psychophysical thresholds. However, there was a high correlation between the data from the tactile and emotional subsets of the questionnaires. These results support the hypothesis that the hyper- and hypo-responsivity to touch, which is sometimes observed in autistic spectrum disorders, is not a perceptual sensory problem, but may probably be emotional in origin. Keywords: Somatosensation, Pacinian channel, non-Pacinian I channel, hyper-sensitivity, hypo-sensitivity, pervasive developmental disorder Introduction Autistic disorder (AD) is among the pervasive developmental disorders (autistic spectrum disor- ders), which include Rett’s disorder, childhood disintegrative disorder, Asperger’s syndrome, and pervasive developmental disorder not otherwise specified (PDD-NOS). AD occurs in about 20 out of 10 000 children, and an autistic spectrum disorder in about 60 out of 10 000 (Fombonne 2003), but the prevalence is increasing probably due to new diagnostic criteria and heightened awareness. According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR, American Psychiatric Association), the diagnostic criteria for AD are (1) qualitative impairment in social interaction, (2) qualitative impairment in commu- nication, (3) restricted repetitive and stereotyped patterns of behavior, interests, and activities, (4) delays or abnormal functioning with onset prior to age 3 years, and (5) the disturbance is not better accounted for by Rett’s disorder or childhood disintegrative disorder. AD is a lifelong disorder Correspondence: B. Gu ¨c ¸lu ¨ , PhD, Assistant Professor, Biomedical Engineering Institute, Bogazici University, Bebek, Istanbul 34342, Turkey. Tel: +90 212 3596413. Fax: +90 212 2575030. E-mail: [email protected]ISSN 0899–0220 print/ISSN 1369–1651 online ß 2007 Informa UK Ltd. DOI: 10.1080/08990220601179418 Somatosens Mot Res Downloaded from informahealthcare.com by Hacettepe Univ. on 03/25/12 For personal use only.
13
Embed
Tactile sensitivity of normal and autistic children
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Somatosensory and Motor Research, March–June 2007; 24(1–2): 21–33
Tactile sensitivity of normal and autistic children
1Biomedical Engineering Institute, Bogazici University, Istanbul, Turkey, 2Child and Adolescent Psychiatry
Department, Autism Clinic, Istanbul Medical School, Istanbul University, Istanbul, Turkey, and 3Child and Adolescent
Psychiatry Department, Hacettepe University, Ankara, Turkey
(Received 4 June 2006; revised 20 October 2006; accepted 11 December 2006)
AbstractMany children with autistic spectrum disorders have unusual reactions to certain sensory stimuli. These reactions vary alonga hyper- to hypo-responsivity continuum. For example, some children overreact to weak sensory input, but others do notrespond negatively to even strong stimuli. It is typically assumed that this deviant responsivity is linked to sensitivity,although the particular stage of sensory processing affected is not known. Psychophysical vibrotactile thresholds of six malechildren (age: 8–12) who were diagnosed to have autistic spectrum disorders and six normal male children (age: 7–11) weremeasured by using a two-alternative forced-choice task. The tactile stimuli were sinusoidal displacements and theywere applied on the terminal phalanx of the left middle finger of each subject. By using a forward-masking paradigm, 40- and250-Hz thresholds of the Pacinian tactile channel and 40-Hz threshold of the Non-Pacinian I tactile channel weredetermined. There was no significant difference between the thresholds of autistic and normal children, and the autisticchildren had the same detection and masking mechanisms as the normal children. The sensory responsivity of each subjectwas tested by clinical questionnaires, which showed again no difference between the two subject groups. Furthermore, nosignificant correlations could be found between the questionnaire data and the psychophysical thresholds. However, therewas a high correlation between the data from the tactile and emotional subsets of the questionnaires. These results supportthe hypothesis that the hyper- and hypo-responsivity to touch, which is sometimes observed in autistic spectrum disorders, isnot a perceptual sensory problem, but may probably be emotional in origin.
The results of the psychiatric assessments are given in columns 3–5. The questionnaire data are given in columns 6–9. For the SensoryProfile questionnaire, the responses from the entire question set, the touch subset, and the emotional/social subset are given separately.HFA¼ high-functioning autism; PDD-NOS¼ pervasive developmental disorder not otherwise specified; ADHD¼ attention-deficit/hyperactivity disorder; IQ¼ Intelligence Quotient; CARS¼Childhood Autism Rating Scale; TI¼Touch Inventory for Elementary-School-Aged Children (raw scores and percentiles); SP¼Sensory Profile questionnaire (positive scores, see text). Empty fields were not measured.
Tactile sensitivity of normal and autistic children 23
required more trials, because their attention some-
times drifted away from the tactile stimulus during
the experiments.
The experiments used the two-interval forced-
choice task, which generates results independent of
the subject’s criterion. The subject’s task was to
decide whether the test stimulus was in the first or
the second interval. The stimulus levels were
changed by using an up–down rule that tracked
thresholds at 0.75 correct probability of detection
(Zwislocki and Relkin 2001). A random selection of
intervals yields 0.50 correct probability, and any bias
for choosing the intervals results in no tracking.
Therefore, if the subject gave inconsistent responses
deliberately, no measurement could be obtained.
This procedure avoids inaccurate threshold measure-
ments. The up–down rule increased the stimulus
level by 1 dB for each incorrect response, and
decreased the stimulus level by 1 dB for three,
not necessarily consecutive, correct responses. The
experiment automatically stopped when the stimulus
level was within �1dB range for the last 20 trials.
The mean intensity level in the final �1dB range was
recorded as the threshold.
First, absolute thresholds of the subjects were
measured separately at 40 and 250Hz without using
a masking stimulus (Figure 1A). Then, the sensitivity
of the P channel was elevated by using a 250-Hz
masking stimulus. During this experiment, the
250-Hz masking stimulus preceded a 250-Hz test
stimulus (Figure 1B). The masking level was
adjusted to yield a shift (5–15 dB) in the threshold
of the P channel for each subject. Finally, the
threshold of the NP I channel was supposedly
measured at 40Hz by using the adjusted masking
level. For this experiment, the 250-Hz masking
stimulus preceded a 40-Hz test stimulus
(Figure 1B). This forward-masking procedure is
required to measure the threshold of the NP I
channel at 40Hz, because at 40Hz, the P channel is
usually more sensitive than the NP I channel (Guclu
and Bolanowski 2005).
Analyses
The psychophysical thresholds are presented in dB
units referenced to 1 mm peak displacement ampli-
tude. The statistical analyses were performed in
MATLAB (The MathWorks, Inc., Natick, MA,
USA). During the psychophysical experiments,
each measurement was repeated 4 times for each
subject. The graphs plot the averages of those
measurements. The error bars are the standard
errors of the mean. Hypothesis testing (two-sample
t-tests) was performed by using the threshold-
displacement values. The correlation coefficients
Figure 1. Stimulus timing diagrams for experiments without masking (A) and with masking (B). The 40- or 250-Hz teststimulus (small burst) occurs either in the first interval (cued by red light) or the second interval (cued by green light).The subject responds while the yellow light is on. In masking experiments, a high-level 250-Hz masking stimulus(large burst) is present in the beginning of each interval.
Tactile sensitivity of normal and autistic children 25
Som
atos
ens
Mot
Res
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Hac
ette
pe U
niv.
on
03/2
5/12
For
pers
onal
use
onl
y.
(Pearson’s) were found between the questionnaire
results and the threshold-displacement values.
Results
Questionnaire data
The results obtained from the Sensory Profile (SP)
and the Touch Inventory (TI) questionnaires are
given in Table I. Both normal and autistic subject
groups have members who scored high or low on
these tests. According to the total questions available
in a given test or the percentile result, the results
were grouped into three equal ranges. The results are
considered to be low if they are in the lowest range,
high if they are in the highest range, and normal if
they are in the middle range.
N1, N2, A3, A4, and A5 have tactually
defensive (hyper-responsive) behaviors according
to the TI (467%). On the other hand, N4, A1,
and A6 show hypo-responsivity according to the
TI (533%). The other subjects have scores
between the scores of the subjects who show
hypo- and hyper-responsivity. It is interesting to
compare the TI data of autistic children with
clinically reported somatosensory problems, which
are listed in the fourth column of Table I. The
psychiatric evaluation revealed that A1, A4, and
A6 did not react normally to pain. A3 was
reported to overreact to various sensory inputs
which include tactile, auditory, and visual stimuli.
The parents of A2 and A5 reported that their
autistic children did not display excessive
responses to somatosensory inputs. The TI data
of A1, A2, A3, and A6 are consistent with the
psychiatric evaluations. However, although A4 was
reported to be hypo-responsive, his TI score is
higher than 85% of the normal population.
Similarly, the parental report of A5 did not suggest
any excessive sensory problems, but A5 scored
very high in the TI test (i.e., tactually defensive
behavior). The normal subjects do not all have
TI scores which lie close to the average of the
normative population. N1 and N2 have very high
scores; N4 has a very low score. These results
show that the tactile defensiveness as measured by
the TI test may vary considerably across the
normal population as well as the autistic popula-
tion. However, there is no statistical difference
between the TI scores of normal and autistic
children (t-test; p¼ 0.986).
The SP test results are not very different from the
TI test results. N1, N2, and A3 have high scores
(483). If only the tactile subset of the SP test is
considered, N1, N2, N3, and A3 have high scores
(416). If only the emotional/social subset of the test
is considered, N1, N2, N3, A1, A3, A4, and A5 have
high scores (416). The results from the tactile subset
of the SP test are generally consistent with the
psychiatric evaluations. Specifically, the subject who
is hyper-sensitive according to the psychiatric evalua-
tion, A3, has the highest score (21). Again, both
normal and autistic subject groups have members
who have high, medium, or low scores; therefore,
there is a large variation among subjects. However,
there is no statistical difference between the SP
scores of normal and autistic children (t-test;
entire SP set: p¼ 0.635, touch subset: p¼ 0.555,
emotional/social subset: p¼ 0.817). It is interesting
to note that some subjects scored high on both
emotional/social and tactile subsets. This suggests
that there may be a correlation between hypo- and
hyper-responsivity and emotional reactions.
Correlation analyses were performed on data
which include all normal and autistic subjects. A
high (r¼ 0.715) and significant ( p¼ 0.009) correla-
tion was found between the results from the tactile
and emotional/social subsets of the SP test
(Figure 2A; n¼ 12). Similarly, a high (r¼ 0.822)
and significant ( p¼ 0.001) correlation was found
between the results from the TI test and the
emotional/social subset of the SP test (Figure 2B;
n¼ 12). This latter correlation is important because,
the TI test was answered by the subjects, but the SP
test was answered by the parents. As expected, there
is also a high (r¼ 0.780) and significant ( p¼ 0.003)
correlation between the results from the TI test and
the tactile subset of the SP test (Figure 2C; n¼ 12).
Finally, a high (r¼ 0.753) and significant ( p¼ 0.005)
correlation was found between the results of the TI
test and the entire SP test (Figure 2D; n¼ 12). These
correlation data show that although the questionnaire
results vary considerably among normal and autistic
subjects, there is a consistent pattern. The subjects
who have more emotional problems according to the
SP test also have more tactile problems according to
the SP test, and display more tactually defensive
behaviors according to the TI test. In addition,
the subjects who have more sensory problems in
general according to the SP test are more tactually
defensive according to the TI test, which also
correlates strongly with the results from the tactile
subset of the SP. Furthermore, the inverse state-
ments are implied because of the given correlations.
For example, the subjects who have very few
emotional problems display hypo-responsive beha-
viors. Similar correlation analyses were performed
also by using the CARS scores for the autistic
subjects. There was significant correlation only
between the results from the emotional/social
subset of the SP test and the CARS scores
(r¼ 0.818; p¼ 0.047), which verifies that emotions
and social interactions are important for determining
the severity of autism.
26 B. Guclu et al.
Som
atos
ens
Mot
Res
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Hac
ette
pe U
niv.
on
03/2
5/12
For
pers
onal
use
onl
y.
Psychophysical data
The psychophysical thresholds of the subjects are
given individually to investigate the detection and the
masking mechanism (Figure 3). The white bars are
the average thresholds obtained without masking
when the stimulus frequency was 40Hz. The 40-Hz
unmasked thresholds of the normal subjects
(Figure 3A) are 6.3, 7.6, 12.8, 9.8, 11.0, and
23.2 dB, respectively, for N1–6. The 40-Hz
unmasked thresholds of the autistic subjects are
20.7, 13.3, 5.8, 10.4, 4.0, and 9.2 dB, respectively,
for A1–6. There is no statistical difference between
the 40-Hz unmasked thresholds of the normal and
autistic children (t-test; p¼ 0.743). The average
40-Hz unmasked threshold among subjects is
14.0 dB for the normal group and 12.5 dB for the
autistic group (Figure 4). The 40-Hz unmasked
thresholds of the normal children were compared to
the data obtained from normal adults (Guclu and
Bolanowski 2005; Kalkanc| and Guclu 2005) and no
significant difference could be found (t-test;
p¼ 0.634). Without masking data (see below), it is
impossible to determine if the 40-Hz unmasked
thresholds are mediated by the P or NP I channel
(Guclu and Bolanowski 2005).
The 250-Hz unmasked thresholds are mediated by
the P channel because of the high sensitivity of this
psychophysical channel at 250Hz (Bolanowski et al.
1988; Gescheider et al. 2001). The 250-Hz
unmasked thresholds are �14.7, �21.6, �11.2,
�16.2, �8.8, and �8.3 dB for N1–6, respectively.
On the other hand, the 250-Hz unmasked thresholds
are �5.8, �10.7, �17.0, �9.4, �15.8, and �10.5 dB
for A1–6, respectively. There is no statistical
difference between the 250-Hz thresholds of
normal and autistic children (t-test; p¼ 0.507). The
average 250-Hz unmasked threshold among subjects
is �12.4 dB for the normal group and �10.7 dB for
the autistic group (Figure 4). No significant differ-
ence was found (t-test; p¼ 0.829) between the
thresholds of normal children and normal adults
(Guclu and Bolanowski 2005; Kalkanc| and Guclu
2005).
In order to determine the threshold of the NP I
channel at 40Hz, the sensitivity of the P channel
should be elevated by masking. Since the masking
effects may change across subjects, the level of the
0
5
10
15
20
25(A) (B)
0 5 10 15 20 25SP emotional/social positive score
0 5 10 15 20 25SP emotional/social positive score
SP
touc
h po
sitiv
e sc
ore
r = 0.715p = 0.009
0
10
20
30
40
50
60
70
TI r
aw sc
ore
r = 0.822p = 0.001
0
10
20
30
40
50
60
70(C)
0 5 10 15 20 25
SP touch positive score
TI r
aw s
core
0
10
20
30
40
50
60
70(D)
TI r
aw s
core
r = 0.780p = 0.003
0 20 40 60 80 100 120
SP overall positive score
r = 0.753p = 0.005
Figure 2. Correlations between the following questionnaire data are presented including all subjects: (A) SP touch subsetscore and SP emotional/social subset score, (B) TI raw score and SP emotional/social subset score, (C) TI raw score and SPtouch subset score, and (D) TI raw score and SP entire set score. Correlation coefficients and p-values are given on the plots.SP¼Sensory Profile questionnaire; TI¼Touch Inventory for Elementary-School-Aged Children.
Tactile sensitivity of normal and autistic children 27
5–15 dB to eliminate its contribution to detection.
The purpose of this procedure was to selectively
measure the threshold of the NP I channel.
Appreciable masking was obtained for each subject.
The threshold shifts obtained by testing a 250-Hz
stimulus applied immediately after the masking
stimulus are 22.8, 29.6, 25.5, 39.8, 10.9, and
36.3 dB for N1–6, respectively. For autistic children,
the masking levels were set lower than the normal
children to minimize attentional lapses which may
occur during the experiment. Therefore, lower
threshold shifts were obtained for the autistic
children. The threshold shifts are 6.5, 14.9, 12.1,
11.4, 14.6, and 10.3 dB for A1–6, respectively. Note,
however, that autistic children displayed a masking
phenomenon indistinguishable from normal
children.
Lastly, the thresholds were measured at 40Hz with
forward masking (gray bars in Figure 3). If masked
thresholds at 40Hz are higher than unmasked
thresholds at 40Hz, the masked thresholds are
considered to be mediated by the NP I channel
and the unmasked thresholds should be mediated by
the P channel (Guclu and Bolanowski 2005).
The 40-Hz masked thresholds of normal children
are 18.5, 24.0, 31.1, 29.9, 17.8, and 39.2 dB for
N1–6, respectively (Figure 3A). For all normal
subjects, the gray bars are significantly higher than
the white bars (t-test; p50.001). This shows that the
white bars are the thresholds of the P channel at
40Hz. The 40-Hz masked thresholds of autistic
children are 26.2, 26.0, 14.6, 15.1, 18.6, and 13.9 dB
for A1–6, respectively (Figure 3B). For all autistic
subjects, the gray bars are significantly higher than
the white bars (t-test; p50.047). This implies that
the detection mechanism found for the normal
subjects is present for the autistic subjects as well.
Furthermore, there is no statistical difference
between the NP I thresholds (gray bars) of normal
subjects and autistic subjects (t-test; p¼ 0.142). The
average NP I thresholds across subjects is 30.1 dB for
the normal group and 20.7 dB for the autistic group
(Figure 4). Although there is about 9.4 dB difference
between the two subject groups, this difference is not
statistically significant because of inter-subject var-
iance. The average NP I threshold of normal adults is
19.6 dB (Guclu and Bolanowski 2005; Kalkanc| andGuclu 2005). The thresholds of normal adults are
significantly lower than the thresholds of normal
children (t-test; p¼ 0.019), but not different than the
thresholds of autistic children (t-test; p¼ 0.613).
−5
5
15
25
35
45
55
−5
5
15
25
35
45
55
(A)
(B)
N1 N2 N3 N4 N5 N6
Thr
esho
ld (
dB r
e 1
µm p
eak)
No masking
Masking
A1 A2 A3 A4 A5 A6
Thr
esho
ld (
dB r
e 1
µm p
eak)
No masking
Masking
Figure 3. Psychophysical thresholds—of normal (A) andautistic (B) children. The white bars are the 40-Hzunmasked thresholds and they are mediated by thePacinian channel (see text). The gray bars are the 40-Hzmasked thresholds and they are assumed to be mediated bythe Non-Pacinian I channel. The error bars are thestandard errors of the mean.
−20
−10
0
10
20
30
40
10 100 1000Frequency (Hz)
Thr
esho
ld (
dB r
e 1
µm p
eak)
Normal P Normal NP IAutistic P Autistic NP I
Figure 4. The average thresholds of the Pacinian andNon-Pacinian I (NP I) channels of normal and autisticchildren, at 40 and 250Hz. Note that the thresholds of theNP I channel could not be measured at 250Hz, becausethe sensitivity of the NP I channel is very low compared tothe Pacinian channel at 250Hz. The error bars are thestandard errors of the mean.
28 B. Guclu et al.
Som
atos
ens
Mot
Res
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Hac
ette
pe U
niv.
on
03/2
5/12
For
pers
onal
use
onl
y.
Correlations between questionnaire data and psycho-
physical data
The analyses in the previous subsections established
that there are no differences between normal and
autistic children regarding the spread of qualitative
sensory problems and the psychophysical tactile
thresholds. However, there seems to be a correlation
between emotional and tactile problems. There may
also be correlations between qualitative sensory
problems and psychophysical thresholds for each
subject group, or when both groups are combined.
This would imply that the qualitative sensory
problems may originate from perceptual deficits,
although this cannot be strictly proven because
correlation does not show a causal relationship.
First, each subject group was treated separately.
Correlations were calculated between the psycho-
physical thresholds (40-Hz Pacinian, 250-Hz
Pacinian, 40-Hz NP I) and questionnaire results
(TI score, SP score, SP touch subset score,
SP emotional/social subset score). Although all
combinations of psychophysical data and
questionnaire data were tested, no significant corre-
lations could be found. For the normal group, the
highest (r¼�0.545; p¼ 0.264) correlation is
between the SP touch subset score and 40-Hz NP I
threshold (Figure 5A; n¼ 6). For the autistic group,
the highest (r¼�0.807; p¼ 0.052) correlation is
between the TI score and 250-Hz Pacinian threshold
(Figure 5B; n¼ 6). Then, all subject data were
pooled and every combination of questionnaire and
psychophysical data was tested. However, no sig-
nificant correlations were found in this analysis
either. If all subjects are included in the analysis,
the highest (r¼�0.547; p¼ 0.066) correlation is
between TI score and 250-Hz Pacinian threshold
(Figure 5C; n¼ 12). Regarding the emotional/social
subset of the SP test, the highest correlation
(r¼�0.152; p¼ 0.638) is for the 40-Hz NP I
thresholds (Figure 5D; n¼ 12) if all subjects are
included. Additionally, no significant correlations
were found between the psychophysical thresholds
and CARS scores for the autistic subjects. These
analyses show that there is probably no causal
0
5
10
15
20
25(A)
(C) (D)
0 20 40 60 80 100
40-Hz NP I threshold (µm)
SP
touc
h po
sitiv
e sc
ore
r = −0.545p = 0.264Normal
0
10
20
30
40
50
60
70(B)
0 0.1 0.2 0.3 0.4 0.5 0.6
250-Hz Pacinian threshold (µm)
TI r
aw s
core
s
r = −0.807p = 0.052Autistic
0
10
20
30
40
50
60
70
0 0.1 0.2 0.3 0.4 0.5 0.6
250-Hz Pacinian threshold (µm)
TI r
aw s
core
s
r = −0.547
p = 0.066
0
5
10
15
20
25
0 20 40 60 80 100
40-Hz NP I threshold (µm)
SP
em
otio
nal/s
ocia
l sco
res
r = −0.152p = 0.638
Figure 5. The highest correlations between the psychophysical and questionnaire data are presented (including all subjectsunless otherwise noted), but there are no significant correlations (see p-values). (A) SP touch subset score and 40-Hz NP Ithreshold, (B) TI raw score and 250-Hz Pacinian threshold, (C) TI raw score and 250-Hz Pacinian threshold, and (D) SPemotional/social subset score and 40-Hz NP I threshold. Correlation coefficients and p-values are given on the plots.SP¼Sensory Profile questionnaire; TI¼Touch Inventory for Elementary-School-Aged Children; NP I¼Non-Pacinian Itactile channel.
Tactile sensitivity of normal and autistic children 29
Som
atos
ens
Mot
Res
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Hac
ette
pe U
niv.
on
03/2
5/12
For
pers
onal
use
onl
y.
relationship between qualitative sensory problems
identified by the questionnaires and psychophysical
thresholds.
Nevertheless, there is some decreasing tendency
(negative slopes) on the plots given in Figure 5.
In other words, as the thresholds increase,
the questionnaire scores decrease, that is, more
hypo-responsivity. As a final attempt to investigate
a possible relationship between questionnaire results
and psychophysical thresholds, the subjects were
arranged into tactile defensive (hyper-responsive:
N1, N2, A3, A4, A5) and not-defensive (hypo-
responsive: N4, A1, A6) groups according to the
analyses presented above. The psychophysical
thresholds of these two groups were compared to
test for the hypothesis that hyper-responsive
subjects have lower thresholds than the thresholds
of hypo-responsive subjects. No statistically signifi-
cant effects could be found for the three thresholds
measured (40-Hz Pacinian: p¼ 0.068; 40-Hz NP I:
p¼ 0.075; 250-Hz Pacinian: p¼ 0.095). Similarly,
the subjects were regrouped into emotional (hyper-
responsive: N1, N2, N3, A1, A3, A4, A5) and non-
emotional (hypo-responsive: N4) groups according
to questionnaire data. Again, no statistically signifi-
cant effects could be found under the same test
hypothesis mentioned above (40-Hz Pacinian:
p¼ 0.574; 40-Hz NP I: p¼ 0.100; 250-Hz
Pacinian: p¼ 0.702). Therefore, there is no evidence
that hyper-responsive subjects have lower thresholds
than the thresholds of hypo-responsive subjects.
Discussion
The most important result in this study is that no
relationship could be found between psychophysical
thresholds and qualitative reports of sensory
problems by the subjects and subjects’ parents.
This result was further supported by the lack of
statistical difference between the thresholds of
normal and autistic children. There was about
9.4 dB difference between the NP I thresholds of
the normal and the autistic group at 40Hz, but this
difference was not statistically significant. If the
means and the variances stay constant while increas-
ing the sample size, such a difference will be expected
to be significant. However, it is also likely that the
mean threshold of the autistic subjects may approach
the mean threshold of the normal subjects if more
subjects are recruited.
There is a link between the emotional and tactile
responses in the questionnaires, which suggests that
the subjects are more likely to have sensory problems
if they have emotional problems. The parsimonious
explanation based on these findings is that the
sensory problems reported in questionnaires or in
clinical evaluation are not based on low-level
perceptual deficits. The correlation between emo-
tional and sensory questionnaire data implies that
there may be higher-order, emotional processes
involved in the sensory problems. Since the sense
of touch has not been studied much in autism as
compared to the modalities of vision and hearing,
this conclusion is rather important.
Although the sample size used in this study is small
(six normal and six autistic children), the psycho-
physical procedures are very robust and have been
frequently used in basic somatosensory research
(Verrillo 1963, 1971; Gescheider et al. 1983, 1994;
Bolanowski et al. 1988; Lamore and Keemink 1988;
Makous et al. 1996; Guclu and Bolanowski 2005).
Two-interval forced-choice task avoids interaction of
subject’s decision criterion and provides objective
threshold measurements even with small sample
sizes. However, the conclusion presented here is
tentative because responses obtained from the
questionnaires are subjective. Nevertheless, such
questionnaires are helpful in the clinical setting to
discriminate developmentally delayed children with
and without tactile defensiveness (Larson 1982),
to discriminate sensory experiences in pervasive
developmental disorders and attention-deficit/
hyperactivity disorder (Ermer and Dunn 1998) and
for many other purposes which do not require much
detailed information on the underlying neural
processes (McCracken 1975; Bauer 1977; Royeen
1986; Dunn 1994). In future studies, emotional
responses of a larger sample of subjects may be
modulated and measured during tactile stimulation
to test the conclusion presented here. However, this
is an ambitious goal because autistic children, even
if they are higher functioning, may probably have
attentional and cognitive difficulties in such complex
tasks.
It is also important to note that the SP and TI tests
have not been done in large populations of normal
and autistic subjects parallel with measuring psycho-
physical thresholds. Future studies may find a link
between the questionnaires and the thresholds,
which would imply that sensory defensiveness
reflects an independent additional impairment.
At this time, however, that statement is speculative,
and the purpose of the current study was not to
support or disprove it. On the other hand, the lack of
statistical difference between the SP scores of normal
and autistic children may be because of the small
sample size.
Cognitive theories on autism
Three non-exclusive theories have been proposed to