You are on page 1of 11

co15 3931. %I 53 co -0.

00
I 1981 P:rgamon Prcu Ltd.

GLOBAL VERSUS LOCAL PROCESSING:


IS THERE A HEMISPHERIC DICHOTOMY?

DAVID B. BOLES*

Department of Psychology, University of Illinois, Champaign, IL 61820, U.S.A

(Accepterf 19 December 1983)

Abstract-A major dichotomy proposed for laterahzed information processing is that the rieht and
left cerebral hemispheres are “global” and “analytic” processors respectively. Here three expe;iments
employed a tachistoscopic recognition paradigm with letter patterns varying in size and composition.
No visual field asymmetries in RT were found in the first two experiments, although statistical power
was high and a Stroop-like effect (“global precedence”) was supported. Experiment 3 resulted in a
right field advantage for vocal naming responses yet no asymmetry for manual responses, suggesting
that both hemispheres can recognize global and local patterns, but that introduction of a laterally
controlled response independently determines asymmetry.

INTRODUCTION
ONE OF the elementary principles proposed to underlie hemispheric differences in function is
that the right hemisphere tends to be a global or holistic processor, while the left hemisphere
is relatively more analytic [S, 151. A common problem in research putting this proposal to
test, however, has been the difficulty of operationalizing the distinction in such a way that
resulting data show independent dissociations between global and analytic processing and
between visual fields (or ears) of presentation. What is needed is a pattern of results in which
global processing is clearly separated from analytic processing, and in which opposite field
asymmetries are demonstrated for each.
Recently MARTIN [16] and SERGENT [24] used adaptations of the visual Stroop task of
NAVON [20] and found results appearing to meet these criteria. In Navon’s task, either (1)
identical letters small in size are arranged in the pattern of a large letter which is either the
same as or different than the small letters; (2) small non-letters make up a large letter; or (3)
small letters are composed into large non-letters. Navon found that large letters are
recognized more quickly than small letters, and that the presence of large letters interferes
with the recognition of small letters if they are different. No such interference was obtained
for small letters upon large letter recognition. This constellation of results is termed “global
precedence”, reflecting quantitatively earlier processing of the “global” or large letters.
While subsequent research indicates that global precedence is not absolute and depends
on such variables as contrast, absolute size, sparsity and distortion [12, 14, 17, 183,
hemispheric logic seems to demand only that there exist conditions which yield global
precedence. MARTIN [16] adopted the paradigm for unilateral presentation to normal
subjects and obtained results generally consistent with global precedence. At the same time,
she reported a general left hemisphere advantage for recognizing small letters, with at least a

*Present address: Department of Psychology, Rensselaer Polytechnic Institute, Troy, NY 12181, U.S.A.

445
446 DAVID B. E~OLES

tendency for a right hemisphere advantage in recognizing large letters. These results are
consistent with a right hemisphere global processor of large visual areas and a left hemisphere
analytic (or “local”) processor of small areas. SERGENT [24] published further results along
these lines. Like Martin, she reported that RTs to dual-level letter patterns produce right or
left hemisphere advantages depending on whether global or local levels are critical to
performance.
While these studies seem to satisfy the criteria of clearly dissociated global and analytic
processing and visual field of presentation, ALIVISATOS and WILDISG [I] were unable to
produce similar results. By having subjects match two letter patterns at either the global or
local level, evidence was found that both hemispheres are capable of processing each. Thus,
although there was more interference of the global on the local level than vice versa,
replicating the global precedence effect, matching at either level did not produce a visual field
difference.
Because the global vs analytic dichotomy is of such potential importance to hemispheric
theory [S] and the global precedence paradigm seems to operationalize it in an elegant
manner (although it can be argued that “local” letters are not typical features of visual
patterns), more data are desirable either to support it or to further the question of
replicability posed by ALIVISATOS and WILDING [l]. Besides addressing the question of
replicability, one of its aims was to follow up an intriguing finding of MARTIN [16].
Interestingly, Martin obtained asymmetries not only in the conflicting conditions but also in
those where either the global or local level was absent. This implies that the global vs analytic
distinction is tied more closely to retinal angle, in which patterns over and under a particular
size are processed by the right and left hemispheres respectively, than to levels of structure, in
which the right hemisphere processes the largest and the left hemisphere the smallest patterns
present. In other words, the results favored a dichotomy tied to absolute size over one tied to
relative size. Thus the three experiments reported here investigate both possibilities.

EXPERIMENT 1

Method
Subjects. Eight males and eight females participated. All were naive undergraduate volunteers from the University
of Texas at Arlington receiving class credit for their participation, and were right-handed by self-report.
Stimuli and apparatus. Three sets ofstimulus cards were prepared using IBM Orator type on white card stock. The
“conflicting” set employed large letters made up ofsmall letters, as illustrated by example in Fig. 1A. Large and small
letters were always different and thus in conflict; all possible conflicting pairings of A, H, I and T were used for a

A A + H
A A T H
A A + H
AAAAAA _----* HHHHHHH
A A H
A A + H
A A + H

A B C
FIG. 1. Examples of stimuli in Experiment I. (A) Conflicting, (B) global neutral, (C) local neutral.
GLOBAL b-ERSUS LOCAL PROCESSING: IS THERE A HEMISPHERIC DICHOTOMY? 447

total of 12 stimuli. Large letters subtended 2.3’ horizontally and 2.9’ vertically (except for the letter “I”. subtending
0.1’ honzontally) and small letters subtended 0.1’ horizontally and 0.3’ vertically. A set of 48 cards was prepared.
with each stimulus appearing twice in each visual field, its center 2.2’ from central fixation.
The “global neutral” set consisted of large letters composed of small plusses. as illustrated in Fig. 1B. The letters
.A. H, I and T were used for a total of four stimuli. Large letters subtended 2.3’ horizontally and 3.7’ vertically
(except “I”, subtending 0.1’ horizontally) and plusses subtended 0.1’ m both dimensions. Forty-right cards were
prepared. each stimulus appearing six times in each visual field. centered 2.2’ from fixation.
The third set was the “local neutral”set, employing small letters arranged in a 1arge”plus” pattern. as illustrated in
Fig. 1C. There were four stimuli, using the letters A. H, I and T. Plusses subtended 7.7’ horizontally and 3.7’
vertically, while small letters subtended 0.1’ horizontally and 0.3’ vertically. Each stimulus appeared six times in
each visual field for a total of 48 cards. Stimuli were centered 2.2’ from fixation.
Stimulus cards were presented in one channel of a Gerbrands 3-channel tachistoscope at a distance of 84 cm. A
second channel was used to present a small central fixation cross. subtending 0.3’ horizontally and vertically. Two
response keys were used, one mapped to the letters “A” and “I” and the other to “H” and “T”. Depressing a key
stopped a Lafayette electronic clock reading to the nearest msec.

Procedure. Although there were only three stimulus sets, the experiment had four conditions since the conflicting
set was presented once requiring recognition of the global letters (the “global conflicting” condition) and once
requiring local letter recognition (the “local conflicting” condition). The four conditions were presented using a
Latin square design, with the I6 subjects filling four different Latin squares.
At the beginning of the experiment the general task was explained and the stimulus ‘key mapping demonstrated.
Then before the first condition, subjects were shown a display of all stimuli in that condition, and practiced pressing
the keys while the experimenter successively pointed to random stimuli on the display. Eye fixation was emphasized,
and instructions were given that RT should be minimized while maximizing accuracy. Depending on the condition,
subjects were told to respond to either the large or small letters and to ignore the irrelevant dimension. Subsequent
conditions were also preceded by this information but not by ad hoc practice of the stimulus/key mapping.
A trial began with the spoken word “ready” as a warning to fixate, and 1 set later the fixation cross was replaced
by a I50 msec stimulus, followed immediately by the reappearing cross. The subject pressed a response key, and the
experimenter recorded which key was pressed and the RT. Keys were arranged with one toward the subject and the
other away, to minimize stimulus-response compatibility effects, and the subject responded using the index fingers
of both hands, one finger per key, with the key/hand assignment balanced over subjects.
At the beginning of each blocked condition, subjects received ten randomly drawn practice trials, followed by 48
randomly ordered experimental trials. Thus within the approximate 1 hr session. subjects received a total of 40
practice and 192 experimental trials,

Results
Median correct RTs were calculated for each visual field in each condition and subjected to
a visual field (left vs right) by level (global vs local) by conflict (neutral vs conflicting)
ANOVA.
The critical visual field x level and visual field x level x conflict interactions were not
significant [F (1, 15) = 2.51 and F (1, 15)~ 1 respectively], as was the visual field main effect
[F (1, 15)~ 11. Figure 2 illustrates relevant data.
Significant effects were those of level [F (1, 15)=24.25, P<O.OOl], conflict
[F (1, 15)=22.24, P<O.OOl] andlevel x conflict [F (1, 15)=5.58, P<O.O5].Thefirst twoare
obvious from Fig. 2; the last replicates the “global precedence” effect, with local judgments
more affected by conflict than were global judgments (increases of 105 vs 48 msec). No other
RT effect was significant.
Percentage errors were subjected to a similar ANOVA. Again there were no effects of visual
field [F (1, 15)~ 11, visual field x level [F (1, 15)~ l] or visual field x level x conflict
[F (1, 15)= 1.231. The only significant main effect was that of level [F (1, 15)=6.46,
P<O.O25], with more errors for local than for global judgments (8.6 vs 4.4%). Visual
field x conflict was the only significant interaction [F (1, 15) = 5.63, P <0.05] : the left visual
field tended to make fewer errors than the right in the neutral conditions (4.8 vs 6.679, while
the reverse was true in the conflicting conditions (8.0 vs 6.6%). An interpretation is not
DAVID B. E~OLES

6001

5501
NEUTRAL CONFLICTING

FIG. 2. Reaction time in Experiment 1. with percentage errors shown by respective conditions
(0 0 = right visual field; A--- D = left visual field).

apparent and the result may reflect a type I error; as will be seen, it did not replicate in
Experiment 2.

Discmsion

In contrast to MARTIN’S [ 161 study, Experiment 1 failed to find differential hemispheric


asymmetry for the global and local levels. But as Fig. 2 indicates, the trend uas in the
predicted direction, so it might be argued that there was a problem of insufficient power in the
experimental design. To address this issue, a power analysis was subsequently conducted
[ 131. It was found that the experiment had ample power to detect the effect size reported by
Martin (power ~0.99 at z=O.Ol to detect an interaction of 86 msec, as nearly as can be
determined from figures for Martin’s neutral and conflicting conditions combined). An
analysis using the conventional figures of power=0.80 and r=0.05 suggests that the
experiment should in fact have detected a true interaction of 37 msec, or in other words,
average opposite visual field asymmetries of about 19 msec for the global and local levels.
Thus on the one hand, Experiment 1 failed to support an effect of a size previously
reported. On the other hand, the possibility remains that a much smaller effect is in fact
present. Experiment 2 was designed to increase the power ofdetecting the critical interaction.
To achieve this the paradigm was transferred to a computer-based system and many more
trials were run per subject.

EXPERIMENT 2
Method
Subjects. Eight male and eight female naive subjects from the University of Oregon participated, all right-handed
by self-report. All received class credit in undergraduate courses for their participation.
Stimuli and apparatus. The stimuli used in Experiment I were adapted here to CRT presentation controlled by
GLOBAL VERSC’S LOCAL PROCESSING: IS THERE A HEMISPHERIC DICHOT041Y? 449

computer. Stimulus sets and pairings remained the same, but stimulus sizes varied slightly. In the “conflicting”
condition, large letters subtended 1.5’ horizontally and 2.6’ vertically (except for “I”, subtending 0.2’ horizontally)
and small letters subtended 0.2’ horizontally and 0.3’ vertically (“1” subtended 0.1’ horizontally). In the “global
neutral” set the corresponding figures were 1.5 and 2.5’ (0.2’) for the large letters and 0.2 and 0.3’ for small “plusses”.
Finally. the “local neutral” dimensions were 2.2 and 2.6’ for large “plusses” and 0.2 and 0.3’ (0.1’) for small letters.
Stimuli were presented on a Hewlett-Packard 13OOA X-Y Display scope at a distance of 50 cm, enforced by the use
of a chin rest. They were centered with the nearest edge 2.7” from fixation, regardless of visual field. A small fixation
cross was again used in central vision.
Trials were controlled by a PDP-15 computer, which also recorded responses from the two keys. As in experiment
1, the keys were arranged with one toward the subject and the other away, but in this case subjects made bimanual
responses, pressing the far key with both index fingers or the near key with both thumbs. Two letters (“A” and “I” YS
“H” and “T”) were mapped to each key. The computer recorded RTs to the nearest msec.
Procedure. As in Experiment 1 the 16 subjects were assigned to four different Latin squares, with the four
conditions blocked. Similar instructions were given, and before each condition the possible stimuli were displayed
on the screen for inspection.
A trial began with the appearance of the central fixation cross for 1 sec. followed by a stimulus for 150 msec. The
program then waited for the subject to respond or until 2 set had passed. The RT and correctness of the response
were displayed for 250 msec, and then a 500-msec delay ensued before the reappearance of the fixation cross.
For each condition one block of 48 randomly ordered practice trials was given followed by four blocks of IS
experimental trials. Thus in the course of about 1 hr a subject received a total of 192 practice and 768 experimental
trials.

Results
Median correct RTs were calculated and entered into a visual field x level x conflict
ANOVA. No effects were found of visual field, visual field x level or visual
field x level x conflict [all F (1, 15)~ 11. Figure 3 illustrates the non-significant three-way
interaction.
The only significant RT effects were those of level [F (1, 15)= 108.66, P<O.OOl] and
conflict [F (1, 15) = 9.37, P <O.Ol]. Somewhat surprisingly, the level x conflict interaction
was not significant [F (1, 15)=2.61], even though in direction, conflict seemingly affected

800
r
t

700-

RT

_-----
89
84

___
NEUTRAL CONFLICTING

FIG. 3. Reaction time in Experiment 2, with percentage errors shown by respective conditions
(0 - - - - 0 = right visual field; n ------A = left visual field).
450 D.AWD B. BOLES

local judgments much more than global judgments (increases of 70 vs 21 msec). However, it
may be that the significance was depressed by the operation ofa practice effect: with the order
of conditions varying over subjects, and the interaction involving only between-condition
differences, considerable variance may have been added to the error term. The problem could
well be more serious than in Experiment 1 (where the interaction was significant) because of
the much greater number of trials here.
One way to attenuate if not eliminate the practice effect statistically is to make the Latin
square rather than the subject the inferential unit. (Within a Latin square, the order of
conditions is balanced.) Employing this logic, the arithmetic interaction was calculated for
each Latin square and its significance assessed by f test. The interaction was indeed
significant [t (3) = 4.35, P < O.OS], replicating the global precedence effect.
Finally, percentage errors were subjected to a similar ANOVA. Visual field had no
influence as a main effect [F (1, l5)< 11, or in interaction with level [F (1, 15) <l] or with
level x conflict [F (1, 15)= 1.071. Unlike experiment 1, no evidence was found of a visual
field x conflict interaction [F (1, 15)< 11.
There were, however, three significant terms in the error analysis: level [F (1, 15) =31.62,
P<O.OOl], conflict [F (1, 15)=24.56, P<O.OOl] and level xconflict [F (1, 15)=13.11,
P < O.OOS].Global judgments produced fewer errors than local judgments (8.7 vs 20.0%), and
the neutral conditions fewer errors than the conflicting conditions (12.4 vs 17.4:/,). The
interaction replicated the global precedence effect, for conflict had no effect on global
judgments, but a substantial effect on local judgments (error changes of -0.2 vs + 10.3%).

Discussion
As in Experiment 1, the critical visual field x level interaction was not significant, failing to
support different roles of the cerebral hemispheres in global vs local processing. Subsequent
power analysis proved revealing: Experiment 2 had 80% power to detect a 21-msec
interaction at x=0.05. A 21-msec resolution corresponds to a mean asymmetry of 11 msec, a
figure below that of the typical published asymmetry using choice RT [e.g. 10,21,25,27,28].
Together the first two studies indicate no hemispheric differentiation of global vs local
levels of processing, and no difference in this regard between tachistoscopic and CRT
presentations. Strictly speaking, however, this conclusion applies only to the “levels of
structure” view of global vs local processing, in which these terms respectively refer to the
largest extent and smaller aspects of the display. It does not apply to the “retinal angle” view
in which such processes are seen as tied to large vs small visual angles instead of their
relationship to one another. This is because of the limited range of angles employed in the first
two experiments (0.3-2.9” vertically). Although the range is similar to Martin’s (0.5-4.1”), it
could be that extending the range (particularly upward) would yield more encouraging signs
of asymmetry.

EXPERIMENT 3
Experiment 3 was designed to test the retinal angle hypothesis. Single letters varying from
0.6 to 11.0” vertically were presented unilaterally, with subjects again pressing RT keys in
response. In addition, a response condition was included in which the letters were to be
named as quickly as possible. The major motivation for this inclusion was that since
Experiments 1 and 2 yielded null visual field results, it was considered desirable to anticipate
this result and to include a manipulation that should yield asymmetry apart from the retinal
GLOBAL VERSUS LOCAL PROCESSING: IS THERE A HEMISPHERIC DICHOTOMY’! 151

angle manipulation. In the absence of other factors influencing asymmetry, naming


responses should theoretically induce a left hemisphere superiority since language is typically
lateralized to that hemisphere [ll]. By this strategy, the sensitivity of the RT paradigm to
asymmetry would be demonstrated and so lend more confidence to a null result for visual
angle.

Method
Subjects. Eight females and four males participated, all right-handed by self-report. Seven were undergraduates
from the University of Oregon who received credit in a class for their participation, while live were from the
University oforegon Cognitive Laboratory subject pool and were paid S3.00 each. All were naive as to the purposes
of the experiment.
Srimuli ad apporarus. Stimuli were the letters A, H, 1 and T presented in four different sizes and two intensities on
the CRT/computer system employed m Experiment 2. The smallest size was 0.3’ horizontal (0.2’ for the letter ‘.r’)
and 0.6’ vertical. The next smallest was 2.8’ horizontal (1.5’ for “I”) and 4.1’ vertical, while the third size was 5.3’
horizontal (2.7” for “I”) and 7.6” vertical. The largest letters subtended 7.7” horizontal (3.9” for “I”) and 11.0’ vertical.
Two intensities were used as a control for the clearly decreasing contrast which accumpamed increasing size. Smce
the letters were composed of small luminous points, and the same number of points was plotted per letter regardless
of size, contrast effectively fell as size passed from small to large. Using two intensities controls for this effect
according to the following logic. First, if visual field asymmetry varies according to letter size, and contrast is
irrelevant to this, as one might expect from the experimental hypothesis, then size but not intensity should affect the
obtained asymmetry. On the other hand, if any observed shift in asymmetry is attributable to contrast and not size,
then intensity should affect this shift, and since contrast covaries with size, an effect of sizeon asymmetry should also
be found. Thus the experimental hypothesis predicts only an effect of size, whereas both size and intensity should
affect asymmetry if their confounding is important. The two intensities used were sufficiently disparate to produce a
substantial main effect, as the results will show.
Letters were presented so that regardless of size, their centers were positioned 6.6’ from central fixation. The
viewing distance was 50 cm, enforced with a chin rest.
In the manual response condition, two keys were used as in experiment 2. In the vocal condition. a
Grasson-Stadtler, model E73OOA-1 voice key was employed, the naming of a letter tripping a relay via a throat
microphone.
Procedure. Half the subjects received the manual condition first followed by the vocal condition, with the reverse
order for the other subjects. Instructions similar to the previous experiments were given, with the exceptions that for
the vocal condition, subjects were told to name the letters out loud as quickly and as accurately as possible and that
their responses would be monitored by intercom from time to time. While this was done on an ad hoc basis to
ascertain that subjects were really producing letter names, no attempt was made to score response accuracy, since
previous experience suggested that naming errors are very rare.
A trial began with the appearance of a small fixation cross for 1 sec. followed by a urnlateral letter for 150 msec.
The computer took the subject’s response or waited 2 sec. then initiated another trial after an additional 750 msec. No
RT feedback was given in this experiment.
At the beginning of the experiment, subjects were shown all the letter stimuli. For each condition one block of 64
practice trials was given, followed by six blocks of experimental trials, Visual fields, sizes and intensities were mixed
randomly within a block. Since there were two blocked response conditions, there was therefore a total of 12s
practice and 768 experimental trials in the course of about 1 hr.

Results
Median RTs were calculated over all trials in a condition. To treat manual and vocal RTs
similarly, errors as well as correct responses were included in these calculations in both.
Median RTs were then entered into a size (vertical subtense 0.6,4.1,7.6, or 11.0”) by intensity
(high vs low) by visual field (left vs right) by response (manual vs vocal) by condition order
(manual/vocal vs vocal/manual) ANOVA.
Visual field had no effect [F (1, lo)= 3.753, nor did it interact with size [F (3, 30)~ l] or
intensity [F (1, 10)~ 11. However, the visual field x response term was significant
[F (1, IO)= 5.61, P<O.OYj. Figure 4 gives the visual field results by size and response, and
illustrates the significant interaction: the right visual field/left hemisphere was faster than
the left field/right hemisphere with vocal responses [601 vs 619 msec; f (11) = 3.38, P <O.Ol]
but not with manual responses [751 vs 757 msec; f (11)=0.68]. Furthermore, a second
D.tvro B. BOLES

lOOOr

I I I I I I
5001 ’
0 2 4 6 8 IO 12
VERTICAL SIZE (DEGREES)

FIG 4. Reaction time in Experiment 3 (0 -~ - -0 = right visual field; a--- n = left visual
field).

analysis to determine whether the presumably higher error contribution in the manual data
might have been responsible for the non-significant visual field result gave no such
indication; with errors removed from the manual RT data, the visual field result became nil
(73 1 vs 73 1 msec). Nor was there an asymmetry in percentage errors for the manual condition
c10.7 vs 10.2”/,; I (11)=0.57]. These results would seem to indicate that the differential
asymmetry in the vocal and manual conditions was real, and not an artifact ofincluding error
RTs in the medians.
There were also three main effects, those of size [F (3, 30)= 12.17, P<O.OOl], response
[F (1, IO)= 15.43, P<O.O05] and intensity [F (1, 10)=63.55, P<O.OOl]. Figure 4 illustrates
the first two, while high-intensity stimuli produced shorter RTs than low-intensity stimuli
(659 vs 705 msec).
Finally, there were two other significant interactions. Intensity interacted with size
[F (3, 30) = 6.24, P < 0.005) and size interacted with response [F (3,30) =4.66, P<O.Ol]. The
latter is apparent in Fig. 4, while the former indicates a greater difference between intensities
for the smallest letters (93 msec) than for the other three sizes (in ascending size: 28,33 and
32 msec).

Discussion
The major findings of Experiment 3 were than (1) there is no dependency of VF asymmetry
on retinal angle, against the absolute size variant of the global vs local hemispheric
hypothesis, and (2) asymmetry did depend on a response factor, vocal naming vs a manual
keypress. The latter revealed a right visual field advantage for naming but not manual
responses, a result predicted by left hemisphere language processing. Its significance implies
that the RT method used in the three experiments was in fact sensitive to hemispheric
asymmetry actually present, and lends credence to the null global vs local outcome. Indeed, it
GLOBAL VW.SUS LOCAL PROCESSING: IS THERE A HE.MlSPHERlC DlCHOTO~fY? 453

graphically demonstrates the equivalence of the hemispheres in processing at these levels:


only with a lateralized response did asymmetry emerge, supporting a special case of the
“relative ability” model of hemispheric function in which both hemispheres perform similar
pattern recognition processes with equal facility.
Secondary results of Experiment 3 included a more-or-less U-shaped function relating RT
to visual angle. A similar result has been reported elsewhere [ 14, 183 and supports the notion
of an intermediate angle optimal for visual recognition.

GENERAL DISCUSSION
Three experiments have failed to reproduce the hemispheric results of MARTIN [16] and
SERGENT [24]. Global and local patterns, whether conceived of in relative or absolute senses,
did not yield visual field asymmetry. Thus the results are in agreement with those of
ALIVISATOSand WILDING [ 11, who likewise failed to find visual differences in the presence of
the global precedence effect.
Interestingly, there appear to be no plausible methodological factors that can account for
the differing findings of the four studies, even if Experiment 3 of this report is discounted on
grounds that it did not employ bi-level stimuli. Of the positive replications, Martin used a
mask after each display while Sergent did not. All replications used card tachistoscopes in at
least one experiment, and had subject samples balanced for sex. Other variables show
overlap between positive and negative replications: (1) both Sergent and I used four-letter
sets; (2) Martin, Sergent, and Alivisatos and Wilding used mixed symmetric and asymmetric
letters; (3) global letter size varied from 1.3 x 1.9 to 2.8 x 4.1” in the positive replications,
while I used sizes varying from 1.5 x 2.5 to 2.7 x 3.7”; (4) local letter size varied from 0.1 x 0.2
to 0.3 x 0.5” in Sergent’s and Martin’s studies respectively, while mine used sizes varying from
0.1 x 0.3 and 0.2 x 0.3”; (5) stimulus durations were either 100 and 150 msec in both positive
and negative replications; (6) 12 and 54 practice trials, and 144 and 270 experimental trials
were given by Martin and Sergent respectively, with the corresponding figures for Alivisatos
and Wilding, and Boles (Experiment 1) being 24 and 40, and 96 and 192; (7) one level of the
stimulus was attended to at a time in all but Sergent’s study; and (8) a two-choice manual
identification response was used by both Sergent and me. In addition, the minimum stimulus
eccentricity, while not strictly overlapping between positive and negative replications, was
rather similar in the studies by Sergent (1.4”) and by Alivisatos and Wilding (2.2”).
Accordingly, it would appear that any attempt to isolate a critical variable accounting for
differences in outcomes between studies would be destined for failure at the onset. There seem
to be none that can be put forward as plausible candidates on the basis of available
information about methodologies.
What, then, can we say about the status of the global precedence paradigm as an
operationalization of the putative global/analytic hemispheric dichotomy? It is clear from
the foregoing discussion that the empirical picture is mixed, and that resolution of the
question is not likely to depend on a single critical study but rather on the building of a
corpus of results from a number of independent, well-controlled studies. In this manner, it is
hoped, it will ultimately become possible to decide in favor of the positive or negative
replications.
Nevertheless, the results of the present experiments fit what might be perceived of as an
increasing trend away from the global-analytic dichotomy as a fundamental principle of
hemispheric function, at least as applied to visual processes specifically (for an opposing view,
454 D.A~ID B. E!OLES

however. see [5]). Thus, although PATTERSOS and BRADSHAW [22] reported congruent
results from face processing. F.AIRWEATHERet al. [93 found that interpretation of such effects
was ambiguous and depended in part on particular stimuli. The predictions afforded by
“same” (global) vs “different” (analytic) responses in visual comparison are frequently not
supported [2, S]. Results from numerous embedded figure, overlapping figure and rod-and-
frame tests (which measure sensitivity to an overall surround vs parts of a display, a relatively
clear global-analytic operationalization) produce almost no support for the hemispheric
dichotomy [3,4.6, 7,233. Given this pattern of null and conflicting results. attention might
best be centered elsewhere in the search for elementary organizing principles of hemispheric
function.

.-Ickno,clerlym~~nts--Iam grateful to Jerry G. Ells and Gerald M. Devins for their comments on the manuscript.

REFERENCES
1. ALIVISATOS.B. and WILDING. J. Hemispheric differences in matching Stroop-type leter stimuli. Corfex 18, i-2 I,
1982.
2. BAGSARA, S.. BOLES. D. B.. SIMION, F. and U,CtILTA,C. Can an analytic/holistic dichotomy explain hemispheric
asymmetries? Corte.r 18, 67-78. 1982.
3. BASSO. A., DERESZI. E., FAGLIONI. P.. SCOTTI, G. and SPtNNLER, H. Neuropsychological evidence for the
existence of cerebral areas critical to the performances of intelligence tasks. Brain 96, 715-728. 1973.
4. BISIACH. E.. CAPITASI. E., NICHELLI. P. and SPINNLER. H. Recognition of overlapping patterns and focal
hemisphere damage. Neurops,wholoyio 14, 375-379, 1976.
5. BRADSHAW.J. L. and NETTLETON, N. C. The nature of hemispheric specialization in man. &ha/.. Brain Sci. 4,
51~91, 1981.
6. COHES. B. D., BERENT,S. and SILVERMAN.A. J. Field-dependence and lateralization of function in the human
brain. Archs :!en. Psxclliaf. 28, 165-167, 1973.
7. CORKIS. S. Hidden-figures-test performance: lasting effects of unilateral penetrating head injury and transient
effects of bilateral cingulotomy. iVruropsyclrolo,qia 17, 585-605. 1979.
s. EGETH. H. and EPSTEIN,J. Differential specialization of thecerebral hemispheres for the perception ofsameness
and difference. Percept. Ps_vchoph_vs. 12, 218-220, 1972.
9. FAIRWEATHER,H.. BRIZZOLARA, D., TABO~~I. P. and U~IILT~~,C. Functional cerebral lateralisation: dichotomy
or plurality? Corrr.~ 18, 51-65. 1982.
10. HAUS. F. Functional dissociation of the hemispheres using fovea1 visual input. Neurops.ycholoc/ia 16,725-733.
1978.
Il. HECAES, H. and ALBERT, M. Human Neurops~cholo,y.v. Wiley, New York, 197X.
12. HOFF&IAN.J. E. Interaction between global and local levels of a form. J. exp. Psycho/.: Hum. Percepr. Per/km.
6, 222-23-1, 1980.
13. KEPPEL. G. Des+r und Ann&is: n Researcher’s Handbook. Prentice-Hall, Englewood Cliffs, NJ, 1973.
14. KINCHLA, R. A. and WOLFE, J. M. The order ofvisual processing: “top-down”. “bottom-up”, or “middle-out”.
Prrcepr. Pswhophys. 25, 225-23 I, 1979.
15. LEVY-AGRESTI,J. and SPERRY, R. W. Differential perceptual capacities in major and minor hemispheres. Proc.
natrt. Acad. Sci. U.S.A., 61, I 15 1, 1968.
16. MARTIS’. iL1. Hemispheric specialization for local and global processing. Neurops.vcholo,yia 17, 33-40. 1979.
17. MARnN, M. Local and global processing: the role of sparsity. Memory Co</ni!. 7, 47G484, 1979.
18. MCLEAS. J. P. Perspectives on the forest and trees: the precedence of parts and wholes in visual information
processing. Disserr. Abs/r. Int. 39(B), 6162-6 163. 1979.
19. MOSCO~IT~H, M. Language and the cerebral hemispheres: reaction-time studies and their implications for
models of cerebral dominance In Communicarion and Affeci: Lunyuuye and Thouyhr. P. PLISER, L. KRAMER
and T. ALLOWAY (Editors). pp. 89-126. Academic Press, New York, 1973.
20. N~vos. D. Forest before trees: the precedence of global features in visual perception. Coclnir. Ps~~hol. 9,
353-383. 1977.
21. NIEDERBUHL,J. and SPRINGER, S. P. Task requirements and hemisphericasymmetry for the processing ofsingle
letters. IVeurops,vcholo,yia 17, 659-692, 1979.
22. PATTERSON,K. and BRADSHAW. J. L. Differential hemispheric mediation of nonverbal visual stimuli. J. exp.
PsychoI.: Hum. Prrcepr. Perform. 1, 246-252, 1975.
23. PITBLADO, C. Visual field differences in perception of the vertical with and without a visible frame of reference.
Neurops~cholo.yia 17, 38 l-392. 1979.
GLOBAL VERSUS LOCAL PROCESSING: IS THERE A HEwSPHER!C DICHOTOMY’? 455

14. SERCENT. J. The cerebral balance of power: confrontation or cooperation’! _f. up. Pqchol.: Hum. Pmqr.
Perform. 8. 153-272. 1982.
25. SMITH. L. C. and MOSCOVITCH. XI. Writing posture. hemispheric control of movement and cerebral dominance
in individuals with inverted and noninverted hand postures during writing. .S~urops~~choloqia 17, 637-644.
1979.
26. STERNBERG. S. The discovery of processing stages: extensions of Donders’ method. .dcra psychoI. 30.27G-3 15,
1969.
27. SUBERI. M. and MC&EVER. W. F. Differential right hemispheric memory storage of emotIonal and non-
emotional faces. .Veuropsycholoqia 15, 757-768. 1977.
28. UWLT~, C. RIZZOLATTI, G., MARZI, C. A., ZA~XONI, G.. FRASZN. C.. COI.ARD-\. R. and BERLLCCHI, G.
Hemispheric differences in the discrimination of line orientation. ?Vrurops?.ziroloc/iu 12, 165-171. 1974.

You might also like