You are on page 1of 15

JOURNALOFN EUROPHYSIOLOGY

Vol. 46, No. 6, December 198 1. Printed in U.S. A.

Tactile Spatial Resolution. I.


Two-Point Discrimination, Gap Detection,
Grating Resolution, and Letter Recognition
KENNETH 0. JOHNSON AND JOHN R. PHILLIPS
Sensory Processes Laboratory, Department of Physiology, University of Melbourne,
Parkville, 3052, Australia

SUMMARY AND CONCLUSIONS 3. In experiment II an attempt was made


to eliminate contact area and overall dimen-
1. The experiments reported here are part sions as variables between the stimuli being
of an investigation of the spatial neural discriminated. Subjects were required to dis-
mechanisms underlying tactile sensation. criminate between stimuli with and without
The first step in such an investigation must gaps. The overall dimensions of the stimuli
be a broad, accurate characterization of tac- were constant and much larger than the
tile spatial discrimination. More particu- gaps. The d’ function was constant for small
larly, the investigation requires specification gaps and then rose steeply (d’ = 0.86 for
of the discrimination behavior that depends gaps, g, ranging from 0.2 to 0.6 mm; d’
strictly on spatial neural mechanisms. The = 4.25(g - 0.55) for g > 0.7 mm). d’ is de-
results of four experimental designs are re- fined here as the separation between two
ported here. No single experiment, taken in unit-normal (a2 = 1 .O) distributions that
isolation, provides a basis for ascertaining would, with an appropriate discrimination
which aspects of spatial discrimination are boundary, produce discrimination judg-
based on spatial information in the neural ments with the same probabilities as those
discharge patterns rather than on intensive observed experimentally. Gap size at thresh-
or temporal information. Taken together, old (P = 0.75) equaled 0.87 mm. Some am-
the results of the four experiments provide biguity arises in the interpretation of these
a consistent basis for examining the neural results. Neurophysiological experiments have
mechanisms underlying tactile spatial dis- shown that one major mechanoreceptive fi-
crimination. ber class is very sensitive to edges. Subjects
2. Experiment I, a modified two-point li- may have discriminated between stimuli on
men test, showed that subjects could reliably the basis of the increased discharge evoked
discriminate between one and two OS-mm- by edges when gaps were present.
diameter points even when there was no sep- 4. Experiment III was designed to hold
aration between the two points. The result overall size, contact area, and edge content
demonstrated a high level of spatial resolu- constant between stimuli. Subjects were re-
tion but discrimination may have been based quired to discriminate between square-wave
on any of a number of neural codes. The gratings (equal gaps and bars) that were
contact area and the overall dimensions of oriented along or across the axis of the fin-
the two stimuli being compared were differ- ger. Subjects could not discriminate grating
ent and, therefore, discrimination may have orientation until the gaps exceeded 0.5 mm
been based on a difference in number of ac- (d’ = 0.0 for gaps < 0.5 mm). For gaps
tive fibers, in total number of impulses, or greater than 0.5 mm (spatial periods > 1.O
in the spatial patterning of the afferent dis- mm) the d’ function rose steeply with a slope
charge evoked by the single- and double- of 4.0 SD units per millimeter gap size (d’
point stimuli. = 4.o(g - 0.5)). Gap size at threshold

0022-3077/8 1 /OOOO-0000$01.25 Copyright 0 198 1 The American Physiological Society 1177

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


M. 0. JOHNSON AND J. R. PHILLIPS

(p = 0.75) equaled 0.84 mm. Neurophysio- stimuli are reported in the two following
logical studies have shown that some pri- papers (23, 24)
mary afferents are sensitive to grating ori- A key issue in the study reported here was
entation. Subjects may have discriminated to design experiments that would allow in-
the stimuli on the basis of differences in dis- ferences concerning the spatial neural mech-
charge rate evoked by the gratings with dif- anisms underlying tactual spatial discrimi-
ferent orientations. nation. The study of spatial discrimination
5. The results of each of the preceding requires the use of two or more stimuli that
experimental designs provided simple mea- are spatially different. The problem is that
sures of spatial resolution but they allowed it is difficult, if not impossible, to devise stim-
the possibility that they were based on non- uli that evoke different spatial patterns of
spatial neural cues. Experiment IV, a tactile neural activity but not different total num-
letter-recognition task, yielded results that bers of impulses or different temporal pat-
could not readily be explained by any non- terns in single fibers (when the stimulus is
spatial neural mechanisms. Subjects’ per- moved laterally). When that is so, how can
formance was as good or better than pre- it be inferred that one type of spatial dis-
dictions based on the results of experiments crimination (e.g., two-point discrimination
II and III, indicating that the results of those or grating resolution) is based on one of these
experiments were not based on some simple cues and not the other? A significant part
nonspatial mechanism. On the basis of in- of the study is concerned with this problem.
formation measures, X0-mm letters were The results of four experimental designs
discriminated at a level midway between are reported here. Each is subject to the dif-
chance and perfect discrimination behavior. ficulties referred to above but, together, they
provide a basis for separating the tactual
INTRODUCTION discrimination results based on spatial neural
mechanisms from those based on intensive
The fingertips and the fovea of the retina mechanisms.’ There are no possible tem-
are analogous in many ways. Each region is poral cues since the stimuli are stationary.
an anatomically specialized part of a two- The first three experiments use experimental
dimensional receptor sheet and each is the designs whose treatment is well developed
behaviorallv preferred site when maximum (16, 17) and they provide simple measures
spatial act&y is required. A fundamental of spatial discrimination. However, each de-
problem in tactile and visual neuroscience sign involves relatively few different stimuli
is to understand how spatial variables of a and training, which means that subjects
stimulus are represented in the afferent re- might have used intensive cues or spatial
sponse (the representation problem) and cues. The fourth design, a letter-recognition
how the afferent response is used in task, makes no attempt to control the inten-
spatial discrimination behavior (the coding sive cues; rather, it employs so many differ-
problem). ent geometrical patterns that it is not rea-
Investigation of the coding problem is sonable to hypothesize that they are classified
ideally based on a detailed map of discrim- ~-
ination behavior across a broad range of * Throughout the paper we distinguish between spa-
tial and intensive mechanisms (and cues, codes, etc.).
stimulus conditions coupled with detailed .A mechanism extracting information from the afferent
characterizations of the discharge patterns discharge patterns is purely intensive if it depends on
evoked by those stimuli. Any hypothesis con- the impulse rates in single fibers but makes no use of
cerning neural coding must then account for the exact locations of the afferent terminals relative to
each other or to the exact timing of impulses relative
a broad range of psychophysical results. This
to each other. A cue or code is purely intensive if it can
paper reports the results of four psycho- be extracted by such a mechanism. A mechanism is
physical experiments designed to measure purely spatial if it depends on the exact locations of
spatial resolution, which employ a variety of active neurons but is unaffected by uniform changes in
spatially configured stimuli applied to the impulse rate and makes no use of the detailed temporal
structure of the impulse patterns in single afferents. A
skin without lateral movement (hereafter code or cue is purely spatial if it can be extracted by
called stationary stimuli). Experiments deal- such a mechanism. A similar definition might be applied
ing with the neural representation of such to temporal mechanisms.

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS

accurately on the basis of some unidimen- bilities (with an appropriate, decision boundary).
sional intensive cue (22). The results of that The response bias is defined as the location of this
experiment, which are complex and yield no boundary (in normal deviates) relative to the
simple measure of spatial discrimination, midpoint between the two distributions used to
define d’ (16).
serve to distinguish which results in the first
In experiments I-III subjects were trained with
three experiments derive from intensive cues feedback until their thresholds (d’ = 1.35) stabi-
and which derive from spatial neural cues. lized: this required only 2 h for each of the subjects
in experiment 1, but about 20 h each for those in
METHODS experiments II and III.
EXPERIMENT I: TWO-POINT DISCRIMINA-
Experiments I-III TION. The stimuli (Fig. 1) consisted of OS-mm-
Each subject sat with his right index finger im- diameter steel pins with flat ends embedded singly
mobilized in a support that allowed stimuli to be or in pairs, near the edge of a large horizontal
applied to the distal finger pad from below. The disk that was rotated to bring the appropriate
stimuli were driven onto the skin by a linear motor stimuli into place. The disk was mounted on a
that allowed either force or displacement to be counterbalanced, pivoting cantilever that allowed
controlled (5). Between stimulus presentations the the whole disk to be raised and lowered. The un-
stimuli fell free of the skin so that the next stim- derside of the disk was supported by rollers. At
ulus could be positioned beneath the finger. The 8-s intervals the linear motor raised the counter-
period between stimuli and the stimulus duration balanced cantilever and disk (rise time, 50 ms;
in each experimental design were adjusted to duration, 3 s), pressing one or two pins onto the
achieve a comfortable stimulus-response rhythm finger pad with a controlled force ( IQ, 30, or 80
for the subject and to allow the experimenter time g). The sequence of two stimuli preceding each
to position the next stimulus, which was done judgment consisted of either S(l) (a pair of pins
manually. The stimulus apparatus and the sub- followed by a single pin) or SC*)(a single pin fol-
ject’s hand were shielded from his view, lowed by a pair). The pairs were aligned along
The experimental designs used in experiments the axis of the finger when brought into place and
I-III employed two possible stimulus sequences, were separated by gaps of 0.0,0.2,0.6, or 1.0 mm.
each containing two stimuli, and allowed two pos- The position of the finger was adjusted so that the
sible responses (16). After every second stimulus two pins of a pair contacted the skin as simulta-
in a train of stimuli, the subject was required to neously as possible. Contact time differences were
make a judgment. The sequence of two stimuli monitored electronically; they were generally less
preceding each judgment was one of two possible than 1 ms and they were never allowed to exceed
stimulus sequences (designated here as S(l) and 2 ms,
SC*)), which were presented in random order and
with equal probability. The order was taken from
a table that had been constructed prior to the
experiment using a random-number generator
The judgments were likewise restricted to one of
two possible responses (designated here as R”“and
R(*)), which were judgments that the two preced-
ing stimuli were presented in the S%equence or
the SC*)sequence. The raw data derived from the
experiments were the frequencies of R(l) and R(*)
following S(l) and SC*). These data are reported
here as sample estimates of the conditional prob-
abilities p( R(*) I St*)) and p( R(*) I SC”), which rep-
resent the frequencies with which the subjects re-
ported that SC*) occurred when, in fact, St*) and
S(l) were presented. All of the psychophysical
measures used here were derived from these two
sample probabilities, which fully define the sub-
jects’ behavior (16). The psychometric functions II I
are direct plots of p( R(*)IS(*)) and p( R(*)IS( ‘) ) ver- FIG. 1. Diagrams showing the relationship between
sus experimental variables. The behavioral sepa- the subjects’ right index finger and the stimuli employed
ration index, d’, is defined as the amount by which to measure tactile spatial resolution. I, two-point dis-
two unit-normal distributions would have to be crimination. 111, gap detection. III, grating resolution.
separated to produce the same response proba- IV, letter recognition (edge of letter shown).

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


K. 0. JOHNSON AND J. R. PHILLIPS

EXPERIMENT II: GAP DETECTION. The stim- formance was evident during the eight 2-h sessions
ulus consisted of a 30-mm-diameter, 2-mm-thick required of each subject.
plastic disk that was pressed edgewise onto the Four sets of 26 sans serif capital letters (3.0,
finger pad (Fig. 1) by the linear motor. Radial 4.5, 5.5, and 8.0 mm high) with 0.5-mm line
slots (gaps), 2 mm deep, were cut at 15mm in- widths raised 1.5 mm above the background (so
tervals around the circumference, leaving a 30- that the skin would not contact the background)
mm segment of the rim with no gap. Two such were machined in plastic. The letter widths were
disks were employed so that segments containing held as closely as possible to 60% of the height
gaps of 0.2,0.4, 0.6,0.8, 1.0, 1.5, or 2.0 mm could although that was not always possible (for the
be presented. The disk was rotated between trials letter I, for example). The letters M and W were
so that the rim segment below the finger contained slightly larger, having widths closer to 80% of the
either a gap or no gap (i.e., the 30-mm segment height. These letters were fixed near the edge of
containing no slot). The rim of the disk was a large horizontal disk that could be rotated to
pressed onto the skin for a period of 3.2 s once bring any letter beneath the subject’s finger (Fig.
every 10 s, reaching a skin displacement of 1,400 1). The subject’s right index finger was fixed in
pm (indentation velocity, 755 pm. s-l; final re- a support that allowed only vertical movements.
action force, 40 g average). The long axis of the At 15-s intervals a light signaled the subject to
contact region between the rim and the skin was depress the letter beneath the finger and to hold
colinear with the axis of the finger and was 2 mm it down for a maximum of 4 s in order to identify
wide and 12 mm long overall. The sequence of the letter. At contact the vertical dimension of
two stimuli preceding each judgment consisted of each letter was oriented along the proximodistal
either S(l) (a stimulus with a gap followed by a axis of the finger pad. The disk was counterbal-
stimulus without a gap) or SC2)(a stimulus with anced so that the embossed letter produced a con-
no gap followed by one with a gap). stant reaction force of 60 g. Disk displacement
EXPERIMENTIII:GRATINGRESOLUTION. The was monitored on an oscilloscope and recorded
on a chart recorder. The mean time to steady
stimuli consisted of square-wave gratings cut into
displacement (from 80 trials selected at random)
the faces of 25-mm square plastic blocks that were
pressed onto the skin (Fig. 1) by the linear motor. was 350 ms. Trials in which the subject failed to
Gratings with spatial periods of 1.0, 1.5, 2.0, 2.5, hold a steady depression, which were rare, were
discounted and run again later.
3.0, 4.0, and 5.0 mm were constructed by cutting
slots (depth, 1.5 mm; width, half the spatial pe- RESULTS
riod) at regular intervals across the face of the
plastic block. The face of the grating lay at right Data were derived from experiments con-
angles to the axis of the linear motor and was ducted on the distal phalangeal pad of the
mounted on a pivot so that it could be rotated right index finger of 14 paid subjects (12
between stimulus presentations. The sequence of females and 2 males between 17 and 25 yr
two stimuli preceding each judgment consisted of of age). Subjects were not selected or re-
either S(‘)(the grating bars aligned along the axis jected on the basis of performance.
of the finger during both the first and second stim-
ulus presentations) or SC2)(the grating bars aligned Experiment I: two-point discrimination
along the finger during the first presentation and This experiment tested the ability of sub-
across the finger during the second presentation). jects to discriminate between single and dou-
Indentation levels of 500,900, and 1,200 pm were
used (rise time, 50 ms; duration, 2.5 s; repetition ble-point stimuli (each point having a OS-
period, 7.0 s), and in some experiments vibration mm diameter). Note that this test differs
at right angles to the face was superimposed on from the classical two-point limen test,
the 900-pm step indentation for a period of 2.0 s which determines the least two-point sepa-
centered in the 2.5-s stimulus period. ration at which the subject feels (has the
subjective impression of) two points. As de-
Experiment IV scribed in the METHODS section, one of two
Experiment IV employed a pattern-identifica- stimulus sequences, S(l) = (two points fol-
tion design using the 26 letters of the English al- lowed by one point) or SC2)= (one point fol-
phabet. Subjects were told that the stimuli were
in the form of embossed, capital letters without lowed by two points), were presented and the
serifs. After each stimulus the subject was re- subject was required to respond with the
quired to specify which letter had been presented. judgment that S(l) or SC2)was presented.
No equivocal responses were allowed. The five These responses were designated as R(l) and
subjects who took part in this experiment received RC2), respectively.
no training or feedback. No improvement in per- The mean performance of four subjects

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS 1181

is illustrated in Fig. 2 where p(Rt2) I S(l)) and p( R12’1 S” ‘) p (R’2’( St2’)


p( RC2)I St2)) are plotted against two-point
separation for each contact force. It is not
obvious that any experimental variable, in-
cluding two-point separation, had a statis-
tically significant influence on performance;
the four subjects performed at a high level
under all conditions. Therefore, the raw data m subject BB
(4,460 stimulus-response pairs), consisting 0 II DF
of the sample probability p(Rt2)) for each
combination of subject (four subjects), se-
quence (SC’)and SC2)),point separation (0.0, Gap width (mm)
0.2,0.6, and 1 .O mm), and contact force ( 10, FIG. 3. Gap detection. Subjects’ ability to discrimi-
30, 80 g), were subjected to a four-way anal- nate between bars (12 x 2 mm) with and without central
ysis of variance. Only three sources of vari- gaps applied to the finger pad (1,400 pm indentation)
ation were significant at the 0.01 level; was assessed by presenting the stimuli in either of two
sequences: So) = a bar with a gap followed by a bar with
namely, sequence (which accounted for no gap, or S2) = a bar with no gap followed by a bar
96.2% of variation) and the interactive terms, with a gap* Subjects were required to judge which se-
sequence x point separation (1.4%) and se- quence had been presented ( Rt2) represents the judg-
quence x subject (0.7%). The effect of force ment that St2) was presented). The left and right sides
of the figure show the probability of Rt2) following So)
was not significant, either alone or as an in- and S(‘) versus the width of the gap. Data are for two .
teractive term. trained subjects.
These results demonstrate the exceptional
resolving capacity of the tactile mechanisms
subserving the skin of the distal pad but they most probably differed in total number of
provide no indication of the nature of those active fibers and total number of action po-
mechanisms. Because of the difference in tentials, as well as in spatial profile. Thus,
overall dimension1 and contact area between central neural mechanisms with no spatial
the single and double-point stimuli, the af- resolving capacity might have discriminated
ferent discharge evoked by the two stimuli between one and two points on the basis of
such intensive cues.
p( R12’( S” ‘) p ( Rt2’1 S12’) Experiment II: gap detection
1.0
p;-
In this experiment the overall dimensions
of the two stimuli being compared were held
constant. Subjects were tested for their abil-
ity to discriminate between stimuli (12 mm
--0.5
x 2 mm bars) with and without central gaps
A log (see Fig. 1 ), which were presented in each
n 309 possible sequential order: S(i) = (gap, no
l 80g
gap), S2) = (no pap, lyP)* The conditional
11.1 ,111
probabilities p( R 2, I SC1) and p( Rt2) I SC2))are
0.5 0.0 0.5 1.0
Gap separating two points (mm) plotted against gap size in Fig. 3 for two
subjects. The data are based on a total of
FIG. 2. Two-point discrimination. Subjects’ ability 3,240 stimulus-response pairs obtained after
to discriminate between one and two OS-mm-diameter training was completed. Unlike the first ex-
points applied to the finger pad was assessed by pre- periment, subjects’ responses were less ac-
senting the stimuli in either of two sequences: S(l) = two
points followed by a single point, or S(‘) = a single point curate at small gap sizes, although they per-
followed by two points. Subjects were required to judge formed more reliably (p = 0.66) than chance
which sequence had been presented (Rt2) represents the at the smallest gap size (g = 0.2 mm). The
judgment that St2) was presented). The left and right threshold, based on the traditional criterion
sides of the figure show the probability of R(2)following
S(l) and Sc2) versus the gap separating the two points. of 75% correct judgments (d’ = 1.35), was
Performance (mean of four subjects) at three contact 0.87 mm. However, the threshold is not a
forces ( 10, 30, and 80 g) is illustrated. sufficient characterization of the results

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


1182 K. 0. JOHNSON AND J. R. PHILLIPS

since the response probabilities were not de- tensive rather than spatial information.
scribed by any simple relationship such as Comparison of the results of this experiment
a cumulative normal function. The manner with the results of experiments III and TV
in which subjects’ behavior varied with gap will indicate that the flat section of the d’
size in the vicinity of threshold is illustrated curve was based on intensive cues, whereas
in Fig. 4. The separation index, d’, is roughly the rising phase was based on spatial infor-
independent of gap size at small gaps and mation in the afferent discharge,
then rises rapidly; the data are approximated
by two linear segments Experiment III: grating resolution
d’ = 0.86 g < 0.7 mm In this experiment the overall dimensions,
the contact area, and the edge content were
= 4.25(g - 0.55) g > 0.7 mm (0 held constant for the two stimuli being dis-
where g represents the gap size in millime- criminated. Subjects were required to judge
ters. The subjects displayed a slight bias to- whether two gratings with the same period,
ward the judgment Rc2) (mean bias = 0. I6 presented sequentially, were presented with
SD units). the same alignment, S(l), or with orthogonal
Although the overall dimensions of the alignments, S2! Two grating-resolution
stimuli with and without gaps were held con- studies were conducted. In the first study the
stant, it cannot be inferred with certainty gratings were stepped onto the skin and he1
that the discrimination performance was stationary, which preferentially ges th
based on the ability of central mechanisms slowly adapting afferent fibers In the
to use spatial information in the afferent second study the gratings were vibrated at
discharge. The gaps introduced edges, which 40 and 200 Hz to maximize the contribu-
probably increased the total afferent dis- tions of the quickly adapting afferents (at
charge (23) and they modified the total con- 40 Hz) and the Pacinian afferents (at
tact area, which may also have altered the 200 Hz).
total afferent discharge (see DISCUSSION The performance of each of the three sub-=
Thus, stimuli with and without gaps might jects who took part in this experiment was
have been discriminated on the basis of in- assessed by computing the conditional prob-
abilities p( R(2)IS(1)) and p( Rc2)IS2)) for each
2.5 of the six grating periods at three skin in-
dentations (500, 90 Y and 1,200 pm). The
a
1 individual data for the three subjects relating
2.0
I B to the stimulus intensity of 900-pm skin in-
dentation are illustrated in Fig. 5, Unlike the
previous two experiments, subjects displayed
1.5 /
the full range of performance from chance
to perfect discrimination over the range of
8 /
the primary stimulus variable (
d I /
riod), As in experiments I and 1
1.0 q / 0 little intersubject variability or response
m
- 8
n- - - - l-
AL- bias (mean bias toward Rc2) for three sub-
0 / jects = 0.026 SD units). In Fig. 6, d’ (mean
0.5 i / for three subjects) is plotted versus grating
/ period for the three indentation levels (data
. BE4 base = 5,200 stimulus-response pairs ob-
/ * DF tained after training was completed). It is
0.0 Lr #in
0.0 0.2 0.4 0.6
I
0.8 1.0
, apparent that increasing the grating period
Gap width (mm)
above about LO mm results in a near-linear
increase in d’, particularly at 900- and 1,200-
FIG. 4. Gap detection-behavioral separation index pm indentation Increasing indentation ap-
(d’) versus gap width. Data points represent the indi-
vidual performances of two subjects. Dashed lines rep- pears to lead to an increase in performance
resent a piecewise linear approximation of their behavior at all grating periods. However, the d’ curves
(see text). intercept the abscissa at a grating period of

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS

p( R’*‘l S” ‘) p( R’*‘l S’*‘) of gratings with periods below about 1.0 mm.
Similarity between the gap detection results
and these results can be seen by expressing
these results in terms of the gap between the
bars rather than the grating periods; then,
the results for 900- and 1,200-pm indenta-
tions are closely approximated by
$’ = 0.0 g < 0.5
= 4.O(g - 0.5) g > 0.5 (2)

Grating period (mm) where g represents the gaps in the gratings


FIG. 5. Grating resolution. Subjects ability to resolve (half the grating period). The threshold eval-
grating stimuli applied to the finger pad was assessed uated at 75% correct judgments (d’ = 1.35)
by applying a grating to the skin twice in succession in for 900 and 1,200 pm corresponded to gaps
either of two sequences: S(l) = the grating bars aligned of 0.92 and 0.84 mm, respectively. The rising
along the finger axis in both the first and second pre- limb of the d’ functions obtained for exper-
sentations, or S(*) = the grating bars aligned along and
then across the finger axis. Subjects were required to iments II and III are very similar (compare
judge which sequence had been presented (R(*) repre- equations 1 and 2).
sents the judgment that S(*)was presented). The left and The grating-resolution experiments were
right sides of the figure show the proability of R(*) fol- repeated at 900-pm indentation, but with
lowing S’) and S(*) versus grating period. The experi-
ment was performed using three levels of skin inden- superimposed vibration normal to the skin
tation (500, 900, and 1,200 ,um). The individual surface at the best tuning frequencies for
performances of three subjects corresponding to 900- cutaneous quickly adapting and Pacinian
pm skin indentation are illustrated. afferents (27). Stimuli designed to engage
the two rapidly adapting mechanoreceptive
about 1.0 mm for all indentation levels, im- afferents had no important effect on the sub-
plying th at no in formation is available to the jects’ performance. Vibration at 200 Hz and
decision process concerning the orien tation IO-pm amplitude had no effect; i.e., the d’
curve was virtually identical to the d’ curve
obtained without vibration. Vibration at 40
Hz, 200 pm and at 200 Hz, 35 ,um raised
the d’ function slightly at 1.0 mm and de-
pressed it at higher frequencies. Under these
conditions (40 Hz, 200 pm and 200 Hz, 35
pm), the d’ functions were closely approxi-
mated by d’ = 0.5 -!- 3.l(g - 0.5) for gaps
ranging from 0.5 to 1.5 mm (i.e., grating
periods from 1.0 to 3.0 mm).
although the overall dimensions, the con-
tact area, and the edge content were held
constant for the two stimuli being discrim-
inated in the grating-orientation test it can-
not be inferred with certainty that the sub-
jects’ responses were based on spatial
structure in the afferent discharge when the
results of this experiment are considered in
isolation. Neurophysiological experiments
1.0 2.0
Grating period (mm) (23) carried out in parallel with the psycho-
physical experiments showed that the dis-
FIG. 6. Grating resolution-behavioral separation charge rates of some single afferents are
index (d’) versus grating period. Each curve represents
the performance (mean of three subjects) for one of the greater when the gratings are oriented along
three indentation levels (500, 900, and 1,200 pm) used the dermal ridges than when they are ori-
in the experiment. ented across the ridges.

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


K. 0. JOHNSON AND J. R. PHILLIPS

Experiment IV: letter recognition formed significantly better than chance for
letters smaller than 3.0 mm. Furthermore,
The problems associated with interpreta- this analysis, based on frequency of correct
tion of the data derived from experiments identifications, underrepresents the subjects’
I-III point to an intrinsic difficulty in the performance, as can be seen in the confusion
general experimental design employed in matrix in Fig. 7. Erroneous judgments were
those experiments. They involve a small predominantly restricted to a small number
number of stimuli presented many times in of alternatives. This consistency is illustrated
succession, usually with training and feed- in Fig. 8B where the curve labeled C rep-
back. When the discriminations are difficult resents the mean frequency of correct judg-
the subject may use neural information (if ments at each letter size over all subjects and
it is available) other than that which the letters; C + 1 represents correct judgments,
psychophysical experiment is designed to C, plus the single most-frequent distractor
test. Thus, rather than try to devise a still for each stimulus (e.g., T for V); and C
more carefully controlled two-sequence x + 2 represents C + 1 plus the second most-
two-response design to eliminate the possi- frequent distractor (e.g., Y for V). Almost
bility of spurious cues, a different design 60% of responses to 3.0-mm letters were con-
strategy was used. If a subject accurately fined to the correct response plus the two
identifies a spatial pattern that is one of most frequent distracters. The logical exten-
many possible patterns without feedback or sion of this approach is an information the-
prior training, then it is unlikely that the oretic analysis (2), which assesses consis-
identifications are based on some nonspatial tency in the stimulus-response matrix rather
aspect of the afferent discharge. That ratio- than accuracy.
nale underlies the letter-recognition experi- In this analysis the subject was regarded
ment. as a communication channel between a
Subjects were required to identify raised sender (the experimenter presenting letters
capital letters by touching them without lat- in random order) and a hypothetical receiver
eral movement, as described in the METHODS listening to the subject’s responses. Before
section. Letters with stroke widths of 0.5 mm the subject’s response the receiver’s uncer-
and heights of 3.0,4.5, 5.5, and 8.0 mm were tainty concerning the stimulus is 4.7 bits
used. Subjects were given no training or (log, 26). After the response the possibilities
feedback. The raw data consisted of a 26 are narrowed (assuming that the receiver
x 26 confusion matrix for each of five sub- knows the subject’s response probabilities)
jects and four letter sizes. Figure 7 illustrates and his uncertainty is reduced. The formulas
a pooled confusion matrix representing the used to calculate the mean information
responses of all subjects to 4.5-mm letters. transfer per response are given in the AP-
The entries represent the number of times PENDIX. The performance measured as in-
that each stimulus evoked each possible re- formation transfer (Fig. SC) is much higher
sponse (e.g., the stimulus letter Y was pre- than that measured simply by the frequency
sented 52 times and evoked the response T of correct judgments (Fig. 8A). That can be
24 times and Y 16 times). seen by calculating the information mea-
The probability of correct identification sures that would have resulted if the errors
(mean over all 26 letters) versus letter height had been randomly (uniformly) distributed
for each of the five subjects is shown in Fig. among the 25 alternatives rather than con-
8A. Three aspects of these data are notable. sistently allocated to a few distracters. The
First, as in the previous experiments, the curve labeled C (mean of five subjects) in
performance of all subjects was similar. Sec- Fig. 80 was constructed by redistributing
ond, the subjects performed at levels well the erroneous responses uniformly and re-
above chance (the probability of correct calculating the information measures. The
identification by chance was p = 0.038; i.e., curve labeled I represents the mean of the
l/26) at all letter heights. Third, the rela- curves illustrated in Fig. 8C. At the smaller
tionship between recognition accuracy and letter sizes the consistent but erroneous re-
letter height was linear over the range tested, sponses accounted for most of the informa-
suggesting that subiects would have ner- tion transfer.

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS 1185

RESPONSE

ABCDEFGHIJKLMNOPQRSTUVWXYZ
A 40 1 .12 4. 1 2 1 52
B . 13414li3:..... 32i:36.:i::::. 52
C 3 522 14 211 1 3 52
D .8 28 2 i: : i i 3 2 52
E 2 i 23 !i i . 1 3 1 i 1 . i 1 1 1 2 52
F i i 130 . il . . li, 5 52
G li,i 5 2. i 1 9 2 3 4 5 i 52
t-l i3.23.6 9 : * 1 3 ii . 1 3 2 2 3 2 52
I . 52 . . . 52
J . 5i . 52
K i : i : ii i i i .l 9 1 5 3 6 52
ZL . . . l . . . . 149 .2 . . 52
L M li : 3 .2; i : : . i : i : : : 52
r N ii:ik:i6.i2.7u ii ' .l 3 2 . . 52
* 0 1715.13.. 3Oi ii . i 52
P 1 .4 i ;i:h32:21i..:::: 52
Q i i il ii:. I 61514 321... 52
R 322 i-41 i . i 8 1 513 2 . . i : l 52
S 28lizi4.:i....i23614 iI... 52
T . . 1 . 2 1. 1 4; 52
U .i. 8. i ii .i : : ij;;l;zzi*;::: 52
V . . . . .2. i . 1 . 1 4 . . 210 4lic 9 . 52
W . . l . . i i 2i 12 224 i 1 52
X . . . .S: 2::3ill.::ii:z 1 12 . 1s 52
Y . . 3i 2 . . 24 . i .16 1 52
Z . .i 1 i, .i 1: ii ..i : 1 : : ii . . : 7 1 30 52
56 69 46 70 79 53 50 39 56 62 33 58 40 58 69 58 19 55 45 92 53 26 43 31 30 62 1352

FIG. 7. Letter recognition-confusion matrix. Subjects were required to identify raised (embossed) capital letters
applied to the finger pad (application force 60 g) at four letter sizes (3.0-, 4.5-, 5.5-, and 8.0-mm letter height).
The confusion matrix illustrated relates to 4.5-mm letters; it was obtained by pooling data for five subjects. Each
entry indicates the number of times that the stimulus Si evoked the response Rj. Stimuli were presented in
pseudorandom order a total of 52 times each.

A close examination of the distracters as- letters (I, J, S, and R) fell outside this simple
sociated with each stimulus letter confirms grouping. I and J were identified incorrectly
the initial assumption that the subjects’ be- on only 1 and 5 of 200 trials, respectively,
havior in the letter-recognition task depends and thus they could not be placed in any
on spatial information in the afferent dis- cluster. The most-frequent distracters for R
charge and not some simple measure of the and S were N, E, A and B, G, respectively.
response like the total impulse rate. The As can be seen in the groupings, the most-
most-frequent distracters associated with frequent distracters are those letters that are
each stimulus letter were obtained by pool- closest in spatial form to the stimulus, in-
ing the confusion matrices (each letter was dicating that the subjects’ responses were
presented 200 times). That analysis is almost based on the spatial form of the neural rep-
completely summarized by the sequential resentation.
groupings BDOQGCE-KZX, HNM WU,
PF-TVY, and AL; the dash is meant to sig- Comparision of performance in
nify that there is significant confusion be- experiments II, III, and IV
tween letters of the two groups, but that the If the results of experiments II, III, and
confusion is not as consistent as the confu- IV all represent the human ability to resolve
sion within groups. With one exception, G, spatial components of tactual stimuli, then
the most-frequent distracters for each letter the results of the three experiments should
are given by the two neighboring letters in be related in a systematic way. Systems with
its group. The two most-frequent distracters limited spatial resolution have long been ap-
for G were 0 and B, which are found in the proximated by mosaics of elements whose
same group but are not neighbors. Only four dimensions represent the smallest resolved

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


1186 K. 0. JOHNSON AND J. IX. PHILLIPS

0001 2 3 4 5 6 7 8

Letter height (mm)


FIG. 8. Letter recognition. A: probability of correct response (mean over 26 letters) versus letter height for five
subjects. Guessing behavior would have resulted in 3.85% correct responses. B: curve C represents the mean
frequency of correct responses over all subjects and letters versus letter height. Curve C + 1 represents the frequency
of responses that were correct or confined to the single most-consistent alternative for each letter (e.g., T for Y).
Curve C + 2 represents C + 1 plus the second most-consistent alternative for each letter (e.g., V for Y). C:
average information conveyed by subjects’ responses for each of the five subjects versus letter height (see text).
D: curve I represents the mean information transferred per response over all subjects versus letter height. Curve
C represents the information measure that would have resulted if there had been no consistency in subjects erroneous
responses (see text).

components (13), such as the gaps in exper- of correct responses. Abscissa values in Fig.
iment II. On this basis gratings should be 9 represent dimensions of the stimuli (i.e.,
resolved when the grating period approxi- gap width, grating period, and letter height).
mates two unit spacings, allowing interunit The dashed curves represent the expected
modulation to represent the peaks and performance in the gap-detection and letter-
troughs. Such a mosaic only begins to rep- recognition experiments relative to the grat-
resent letters at somewhat larger dimen- ing-resolution results. The abscissa values of
sions. A 3 x 3 array may allow I, .I, and L the dashed curves are scaled in the ratios 1:2
to be represented effectively but most letters and 5:2. As can be seen, the results of the
require at least 5 x 5 resolution units to re- three experiments lie in approximately the
move geometric ambiguities (e.g., consider expected ratios. It might be argued that the
a dot-matrix display of the letters B, E, M, letter-recognition results should have been
R, S, and W). Based on this simple analysis better than the 5:2 ratio because some letters
it would be expected that the gaps, grating can be resolved by 3 x 3 arrays. As dem-
periods, and letter heights would fall ap- onstrated in Fig. 8, the probability of correct
proximately in the ratio 1:2:5 when com- identification underestimates subjects’ per-
pared at a common level of performance. formance in experiment IV. When the re-
The results of the three experiments are rep- sults of all three experiments are compared
resented in Fig. 9 in terms of the frequencies using an information theoretic measure (bits

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS 1187

1.0 3
&
0.8 g
G
0.6

CE
-.-.- o, 0.6
-
%

0 1 2 3 4 5 6 7 8 9 10
Stimulus dimension (mm)
( gap width, grating period, letter height )
FIG. 9. Comparison of the results of experiments II, III, and IV. Points and solid lines represent results of the
three experiments (measured as probability of correct response) plotted versus the stimulus dimensions: i.e., gap
width (II), grating period (III), and letter height (IV). The dashed curves repeat results of experiment III with
the stimulus dimensions scaled in the ratio 1:2 and 5:2. They represent results for gap detection and letter recognition
that would be expected if the gaps, grating periods, and letter heights had fallen in the ratio 1:2:5 at comparable
levels of performance (see text).

per response) the letter-recognition results code), or 2) it may be based .on some dif-
lie in the range 3:2-4:2 relative to the grat- ference in the magnitudes of the responses
ing-resolution results. On this basis, the let- to the stimuli (i.e., an intensive code). In-
ter-recognition results are somewhat better cluded in this latter category is any aspect
than predicted. of the afferent discharge that might be used
The consistency of the results from ex- by a central mechanism with little or no spa-
periments II, III, and IV indicate that they tial resolution (e.g., the total number of ac-
all measure that facet of tactile spatial res- tive fibers, the total number of action poten-
olution that is based on spatial modulation tials, the peak discharge rate, etc.).
of the afferent discharge. No other hypoth- The experiments reported in the RESULTS
esis accounts for the consistency of the data. attempt to define which coding procedure is
actually employed under various conditions
DISCUSSION and to measure the limit of resolution when
a spatial code is used. Taken together, the
The psychophysical experiments reported four experiments provide a broad, consistent
in this paper were conducted in order to pro- description of the spatial resolving capacity
vide behavioral data necessary for testing of the tactile system for stationary stimuli.
hypotheses concerning neural coding of the In experiment I subjects discriminated
spatial details of tactile stimuli. When mov- two OS-mm-diameter points applied to the
ing stimuli are employed a large number of finger pad from one point with near cer-
hypotheses are available ( 18), partly because tainty, even when there was no separation
of the possibility of coding spatial informa- between the two points. This result is similar
tion in the temporal discharge profiles of to Friedline’s (8) observations made on the
mechanoreceptive afferents. The use of sta- forearm (see later); both results imply that
tionary stimuli, as in this study, limits the the classical two-point limen is not a mea-
number of possible coding procedures to two sure of the limits of spatial resolution (dis-
main types: I) discrimination between spa- cussed later). However, little can be inferred
tially different stimuli may be based on the from the results of experiment I concerning
difference in form of the spatial neural pro- neural mechanisms when the results are con-
files representing the stimuli (i.e., a spatial sidered in isolation. The results of experi-

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


K. 0. JOHNSON AND J. R. PHILLIPS

ments II, III, and IV suggest that the sub- explanations based primarily on intensive
jects might have been able to discriminate cues are not tenable. Also, while the effect
two OS-mm points from one point on the of orientation between the bars and the skin
basis of a spatial code but with extreme dif- ridges on the discharge rates is significant
ficulty; the ease with which they made the (23), the circular patterning of the ridges on
discriminations suggests that they used in- the human fingertip provide a mitigating
tensive cues. The tactile system is very sen- effect. Thus, the evidence indicates that the
sitive to differences in area and length (19, grating-resolution data provide a measure
3 1), which may well be based on intensive of the component of spatial tactual discrim-
information in the manner proposed by Hol- ination that is based on spatial structure in
den (14) to explain the classical two-point the afferent discharge.
limen. The data indicate that for grating periods
The second experiment is, we believe, an less than 1.0 mm no spatial neural infor-
example of two neural codes underlying dis- mation reaches the final discrimination
crimination behavior in one task. At 0.2-, mechanisms. At spatial periods greater than
0.4-, and 0.6-mm gaps the subjects’ discrim- 1.0 mm the behavioral separation index, d’,
ination performance was constant (d’ = 0.86) rose rapidly, indicating the rapid develop-
and significantly better than chance behav- ment of spatial information in the afferent
ior. The following paper (23) shows that the discharge, which was preserved by the cen-
slowly adapting afferents are very sensitive tral pathways.
to edges, which provides at least one expla-
nation for this behavior; that is, that the Comparison of visual and tactual
edges produced increased discharge rates, spatial resolution
which were used to discriminate stimuli with The letter-recognition results provide a
gaps from stimuli without gaps. Increases in basis for comparing visual and tactual spa-
gap size above 0.6 mm produced improved tial discrimination. Confusion matrices rep-
performance, which might be accounted for resenting the performance of subjects in let-
by some nonspatial mechanism but the sim- ter-recognition tasks provide detailed maps
ilarity between the gap-detection and grat- of spatial-discrimination behavior over a
ing-resolution results suggests that they were wide range of geometric forms and they have
both based on the same mechanism. been obtained for vision (11, 12, 30) and
The grating-resolution results are, we be- touch (6, 21; this study). Unless the same
lieve, the key results. They were derived fonts are used and the confusions are in-
from a two-sequence by two-response ex- duced in the same way, it is difficult to com-
perimental design, which is supported by a pare data. Confusion in letter recognition
well-developed theoretical framework ( 16) has been induced bY reducing the let ter size
and they yielded simple measures of spatial (12; this study ), the exposure time (6, 11,
resolution. Of the experiments using this 30), and by presenting blurred images (1).
design (I, II, and III), the grating-resolution The most similar published visual and tac-
task included the most extensive controls for tual data are those of Gilmore et al. (11)
eliminating intensive cues. Furthermore, and Craig (6) who used dot arrays and short
comparison of the results of experiments II, exposure times. The results of their experi-
III, and IV confirmed the hypothesis that ments were very similar, at least as measured
the results of experiment III were based on by the product-moment correlation of fre-
spatial neural cues. quencies of correct judgments (r = 0.88).
One possible ambiguity in the interpre- The visual letter-recognition study that is
tation of the grating-resolution results is that most like the one presented here and that
the change in grating orientation relative to provides a basis for comparing spatial res-
the skin ridges caused differences in the af- olution in the two systems is the study by
ferent impulse rates, which may have been Hartridge and Owen (12) who used visual
used in the discrimination task. The primary letters with heights ranging from 1.4 to 3.4’
.
evidence that this did not occur was the con- of arc. A scaling factor of 2.0-2.5 mm/min
sistency between the grating-resolution and of arc is required to obtain overlap between
letter-recognition results (see Fig. 9) where their psychometric function (percent correct

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS 1189

versus letter height) and the ones presented clear what is being measured at any one lo-
here (Fig. 8A). The imprecision in the ap- cation. For example, it does not represent a
propriate scaling factor arises because the subject’s ability to discriminate two points
visual psychometric function was steeper from one point. That was demonstrated by
than the tactual function. Work of our own Friedline (8) when she showed that subjects
(unpublished observations) using visual and could reliably discriminate two points sep-
tactual letter-recognition tests in single sub- arated by 2 mm from two points with no
jects verified this comparison. All the data separation on the forearm where the clas-
suggest that insofar as spatial neural mech- sical two-point limen is 20 mm. The results
anisms are concerned, 2.0-2.5 mm on the of experiment I showed that on the fingertip,
skin of the distal phalangeal pad is equiva- as on the forearm, subjects’ abilities to make
lent to 1.O’ of arc in the region of maximum discriminative judgments involving punctate
visual acuity. stimuli are much finer than is indicated by
This dimensional equivalence corresponds the two-point limen. Thus, in the classical
closely to the receptor spacings in the skin experiment subjects do not say that the stim-
and the retina, implying that the relation- ulus feels like two points until the separation
ships between receptor spacing and resolu- far exceeds a level at which they are objec-
tion limits are also similar in the two sys- tively able to discriminate two points from
tems. The mean center-to-center spacing of one point. If that is so, subjects should be
cones in the fovea is approximately 2.0 pm able to subclassify and discriminate two-
(25) which corresponds approximately to point separations between the absolute
0.45’ of arc. The innervation density for the threshold (which was not reached in the ex-
distal finger pad is much more difficult to periment reported here-the 0.5-mm pins
assess but two studies in man and monkey were too coarse) and the classical two-point
(7, 15) estimate the mean center-to-center limen. Indeed, the earliest investigations of
spacing for both slowly and quickly adapting tactual discrimination using punctate stimuli
afferents at aproximately 1.0 mm, which (10, 20, 28) showed that subjects could re-
yields a ratio of 2.2 mm/min arc (i.e., 1.0 liably categorize two-point separations be-
mm/0.45’ arc). The letter heights yielding low the classical two-point limen.
50% correct responses (approximately 2.0’ What do these observations mean in terms
of arc in the visual task (12) and 5.0 mm of the underlying mechanisms? We believe
in the tactual task, see Fig. 8) span approx- that this question can be answered by re-
imately five receptors in each system. Thus, phrasing the hypotheses of early investiga-
in both systems the limits of spatial resolu- tors such as Titchener (29) and Boring (3)
tion and of pattern recognition seem to be in terms of the likely central neural repre-
set by the peripheral receptor spacing in the sentation of two points (9) and the results
regions of highest acuity; thus, those path- of this study. To quote Boring (Ref. 4, p.
ways that lead to spatial pattern recognition 482), “It means that there is a series of dif-
must be characterized by optimum or near- ferentiated perceptual patterns, that the cri-
optimum spatial processing. terion for two must be placed at some critical
point in the series, and that the meaning of
Classical two-point limen the judgment two is indeterminate unless the
It is not clear how the spatial discrimi- criterion has been established.” Paralleling
nation results reported here are related to this series of perceptual patterns is a series
the classical two-point limen devised by of neural patterns, all different in spatial
Weber in the 19th century (26). In the clas- form and intensity. At the lowest discrimi-
sical test the separation of a pair of blunt nable levels the spatial differences are al-
compass points is adjusted until the subject most certainly too small to be discriminated;
reports that he feels two points; that sepa- therefore, the subject uses some intensive
ration is usually 2-3 mm on the fingertips, aspect of the response and places his crite-
20 mm on the forearm, and 40 mm on the rion somewhere along that dimension. Then,
back (32). It appears that the two-point li- at separations of the order of 0.5 mm on the
men is correlated with the relative spatial fingertip he begins to resolve the spatial dif-
acuities of these skin regions but it is not ferences. Finally, at 2- to 3-mm separation,

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


K. 0. JOHNSON AND J. R. PHILLIPS

which is well above his limit of resolution, where p(Si I RJ represents the probability that the
the spatial neural patterns become suffi- ith stimulus &as presented given that the subject
ciently bimodal to satisfy the subject’s cri- gave the jth response, which is estimated by nij/
terion for the statement “the stimulus feels Ilj = entry in ith row and jth column of the
(Ilij

like (evokes the subjective sensation of) two Confusion matrix, nj = jth Column total)* The
uncertainty reduction following Rj (i.e., 4.7
points.” Thus, the two-point limen should be
- HWR,)) re p resents the information transfer
regarded as a boundary on a categorical
(1)
scale and not as a threshold of spatial acuity.
I = 5 p(Rj)(4.7 - H(S 1R.i))
APPENDIX i=l

Information theory measures used in = 4.7 - 2 El H(SIRj)


letter-recogn ition analysis i=l N (A3
The subject is regarded as a less-than-re-
liable communication channel between the in which p(Rj) represents the proability of Rj and
experimenter who is selecting letters for pre- N represents the total number of stimulus-re-
sentation to the communication channel and sponse pairs.
a receiver who is listening to the subject’s
responses. Before the subject responds, the ACKNOWLEDGMENTS
receiver only knows that the letters are
equally likely and his uncertainty concerning We thank Lyn Wood and Gloria Kruba who prepared
the typescript.
which letter was selected is given by H(S)
The work was supported by National Health and
= log, 26 (see Ref. 2 j. Following a particular Medical Research Council of Australia grants.
response, Rj, the receiver’s uncertainty con-
cerning any particular stimulus, Present address of IS. 0. Johnson: Dept. of Physiol-
by H(Si I Ri) = - log, p(Si I Rj). His average ogy, The Johns Hopkins School of Medicine, Baltimore,
MD 21205.
uncertainty concerning all stimuli is given
Present address of J* R. Phillips: Physiological Lab-
by oratory, Downing St., Cambridge CB2 3EG, England.
H(S I Rj) = 5 p(Si I Ri) log, p(Si I Rj)
i=l

Received 23 December 1980; accepted in final form


(AI) 30 June 1981.

REFERENCES

1. APKARIAN-STIELAU, P. ANDLOOMIS, J.M.A com- 9. GARDNER, E.P. ANDSPENCER, W.A.Sensoryfun-


parison of tactile and blurred visual form percep- neling. %I[. Cortical neuronal representation of pat-
tion. Bercept. Psychophys. 18: 362-368, 1975. terned cutaneous stimuli. 9. Neurophysiol. 35: 964-
2. ASH, R. Information Theory. New York: Wiley, 977, 1972.
1965. 10. GATES, E. J. The determination of the limens of
3. BORING, E. 6. Cutaneous sensation after nerve single and dual impression by the method of con-
division. Q. J. Exp. Psychol. 10: 1-95, 19 16. stant stimuli. Am. J. Psychol. 26: 152-l 57, 19 15.
4. BORING, E. G. Sensation and Perception in the II. GILMORE,G.C.'HERSH, H., CARAMAZZA, A., AND
History qf Experimental Psychology. New York: GRIFFIN, J. Multidimensional letter similarity de-
Appleton, 1942. rived from recognition errors. Percept. Psychophys.
5. CHUBBUCK, J. G. Small motion biological stimu- 25: 425-431, 1979.
lator. Appl. Phys. Lab. Tech. Digest May-June, 19- 12. HARTRIDGE, H. ANDOWEN, H. B. Test types. Br.
23, 1966. J. Ophthalmol. 6: 543-549, 1922.
6. CRAIG, J. C. A confusion matrix for tactually pre- 13. HELMHOLTZ, H. Physiological Optics, translated
sented letters. Percept. Psychophys. 26: 409-411, by J. P. C Southall. Menasha, WI: Opt. Sot. Am.,
1979. 1924, vol. 1.
7. DARIAN-SMITH, I. AND KENINS, P. Innervation 14. HOLDEN, A. L. Receptive fields and the two-point
density of mechanoreceptive fibres supplying gla- threshold. J. Theoret. Biol. 11: 22-29, 1966.
brous skin of the monkey’s index finger. J. Physiol. 15. JOHANSSON, R.S. ANDVALLBO, A. B.Tactile sen-
London 309: 147-155, 1980. sibility in the human hand: relative and absolute
8. FRIEDLINE, C. L. Discrimination of cutaneous pat- densities of four types of mechanoreceptive units in
terns below the two-point limen. Am. 9. Psychol. glabrous skin. J. Physiol. London 286: 283-300,
29: 400-419, 1918. 1979.

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.


SPATIAL TOUCH: PSYCHOPHYSICS 1191

16. JOHNSON, K. 0. Sensory discrimination: decision 25. PIRENNE, M. H. Visual acuity. In: The Eye, edited
process. 1. Neurophysiol. 43: 177 1- 1792, 1980. by H. Davson, London: Academic, 1962, vol. 2, p.
17. JOHNSON, K. 0. Sensory discrimination: neural 175-196.
processes preceding discrimination decision. J. Neee- 26. Ross, H. E. AND MURRAY, D. .?. E. PI. Webec The
rophysiol. 43: 1793-1815, 1980. Sense of Touch. London: Academic, 1978.
JOHNSON, K. 0. AND LAMB, G. D. Neural mech- 27 TALBOT, W. H., DARIAN-SMITH, I., KORNHUBER,
anisms of spatial tactile discrimination: neural pat- . H. H., AND MOUNTCASTLE, V. B. The sense of
terns evoked by Braille-like dot patterns in the mon-
flutter-vibration: comparisons of the human capac-
key. J. Physiol. London 3 10: 117- 144, 198 1.
ity with response patterns of mechanoreceptive af-
JONES, M. B. AND VIERCK, C. J. Length discrim- ferents from the monkey hand. J. Neurophysiol. 3 1:
ination on the skin. Arue. J. Psychol. 86: 49-60,
301-334, 1968.
1973.
28. TAWNEY, G. The perception of two points not the
20. LASKI, E. DE. On perceptive forms below the level
space threshold. Psychol. Rev. 2: 585-593, 1895.
of the two-point limen. Am. J. Psychol. 27: 569--
571, 1916. 29. TITCHENER, E. G. On ethnological tests of sensa-
21. LOOMIS, J. M. Tactile letter recognition under dif- tion and perception with special reference to the
ferent modes of stimulus presentation. Percept. Psy- tests of color vision and tactile discrimination de..
chophys. 16: 401-408, 1974. scribed in the reports of the Cambridge Anthro-
22. MILLER, C. A. The magical number seven, plus or pological Expedition to Torres Straits. Proc. Am.
minus two; some limits on our capacity for pro- Phil. Sot. 55: 204-236, 1916.
cessing information. Psychol. Rev. 63: 8 I-97, 1956. 30. TOWNSEND, J. T. Alphabetic confusion: a test of
23. PHILLIPS, 9. R. AND JOHNSON, K. 0. Tactile spatial models for individuals Percept. Psychophys. 9: 449--
resolution. IL. Neural representation of bars, edges, 454, 1971.
and gratings in monkey primary afferents. J. Neu- 3 1. VIERCK, C. J. AND JONES, M. B. Size discrimina-
rophysiol. 46: 1192-1203, 1981. tion on the skin. Science 163: 488-489, 1968.
24. PHILLIPS, J. R. AND JOHNSON, K. 0. Tactile spatial 32. WEINSTEIN, S. Intensive and extensive aspects of
resolution. III. A continuum mechanics model of tactile sensitivity as a function of body-part, sex and
skin predicting mechanoreceptor responses to bars, laterality. In: The Skin Senses, edited by D. R.
edges, and gratings. J. Neurophysiol. 46: 1204- Kenshalo. Springfield, IL: Thomas, 1968, p. 195-
1225, 1981. 218.

Downloaded from journals.physiology.org/journal/jn at UB Wuppertal (132.195.142.062) on December 20, 2023.

You might also like