hemispheric asymmetry in use of semantic category information

7
wl2&393?,8153.00-0.00 c_ 1984 Porgmon Pzs Ltd. HEMISPHERIC ASYMMETRY IN USE OF SEMANTIC CATEGORY INFORMATION DAVID HINES, PATRICIA K. SAWYER, JASON DURA, JAMES GILCHRIST and MARY CZERWINSKI Department of Psychological Science. Ball State University. Muncie. IN 47306. U.S.A (Accepted 29 March 1984) Abstract-When two pictures of common objects were presented sequentially. the second was named more quickly if both were members of the same semantic category. This semantic priming effect occurred only when both pictures went directly to the left hemisphere. If the target or prime stimulus was presented to the left visual field-right hemisphere, no priming effect was observed. These results suggest that semantic category information is activated and used by the left hemisphere of the brain. INTRODUCTION A NUMBER of studies have indicated that the left hemisphere is more efficient in recognition [4,9] and short-term memory [6] for verbal stimuli than the right hemisphere. The present study focuses on a deeper level of processing of verbal information: use of semantic category information in long-term memory. Visual information can be presented directly to either the left or right hemisphere by means of tachistoscopic presentation to the left or right visual half-field, since a stimulus in either visual half-field is transmitted to the contralateral hemisphere via the optic nerve. Recent evidence suggests that, at least for briefly presented visual stimuli, the early stages of processing are carried out by the hemisphere to which the information is initially transmitted [4, 9, 111. Thus, a visual stimulus presented briefly to the left visual field will be initially processed in the right hemisphere while a stimulus to the right visual field will be initially processed in the left hemisphere. The present study used a recently-developed variation of the semantic priming task [ 11. In this semantic priming procedure two visual stimuli are presented about 500 msec apart and the subject is required to name the second. The first stimulus presented “activates”a semantic category (or categories) in long-term memory storage. The second stimulus is then processed more quickly when it is a member ofthe activated semantic category [2]. In the present study, each initial (priming) stimulus was presented at a duration just below the threshold duration at which a particular subject named the stimulus. It was hypothesized that the below- threshold presentation would minimize interhemispheric transfer of the priming stimulus when it was presented to one of the visual half-fields. Presenting the priming stimulus slightly below naming threshold does not seem to alter the magnitude of the semantic priming effect [8], particularly when a cognitive masking paradigm is used [3]. The semantic priming task was used in the present study to test the use of semantic category information when stimuli were initially available to the left or right hemisphere. Both the prime and target stimuli were presented at three different locations: to the left visual 427

Upload: david-hines

Post on 25-Aug-2016

217 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Hemispheric asymmetry in use of semantic category information

wl2&393?,8153.00-0.00 c_ 1984 Porgmon Pzs Ltd.

HEMISPHERIC ASYMMETRY IN USE OF SEMANTIC CATEGORY INFORMATION

DAVID HINES, PATRICIA K. SAWYER, JASON DURA, JAMES GILCHRIST and MARY CZERWINSKI

Department of Psychological Science. Ball State University. Muncie. IN 47306. U.S.A

(Accepted 29 March 1984)

Abstract-When two pictures of common objects were presented sequentially. the second was named more quickly if both were members of the same semantic category. This semantic priming effect occurred only when both pictures went directly to the left hemisphere. If the target or prime stimulus was presented to the left visual field-right hemisphere, no priming effect was observed. These results suggest that semantic category information is activated and used by the left hemisphere of the brain.

INTRODUCTION

A NUMBER of studies have indicated that the left hemisphere is more efficient in recognition [4,9] and short-term memory [6] for verbal stimuli than the right hemisphere. The present study focuses on a deeper level of processing of verbal information: use of semantic category information in long-term memory.

Visual information can be presented directly to either the left or right hemisphere by means of tachistoscopic presentation to the left or right visual half-field, since a stimulus in either visual half-field is transmitted to the contralateral hemisphere via the optic nerve. Recent evidence suggests that, at least for briefly presented visual stimuli, the early stages of processing are carried out by the hemisphere to which the information is initially transmitted [4, 9, 111. Thus, a visual stimulus presented briefly to the left visual field will be initially processed in the right hemisphere while a stimulus to the right visual field will be initially processed in the left hemisphere.

The present study used a recently-developed variation of the semantic priming task [ 11. In this semantic priming procedure two visual stimuli are presented about 500 msec apart and the subject is required to name the second. The first stimulus presented “activates”a semantic category (or categories) in long-term memory storage. The second stimulus is then processed more quickly when it is a member ofthe activated semantic category [2]. In the present study, each initial (priming) stimulus was presented at a duration just below the threshold duration at which a particular subject named the stimulus. It was hypothesized that the below- threshold presentation would minimize interhemispheric transfer of the priming stimulus when it was presented to one of the visual half-fields. Presenting the priming stimulus slightly below naming threshold does not seem to alter the magnitude of the semantic priming effect [8], particularly when a cognitive masking paradigm is used [3].

The semantic priming task was used in the present study to test the use of semantic category information when stimuli were initially available to the left or right hemisphere. Both the prime and target stimuli were presented at three different locations: to the left visual

427

Page 2: Hemispheric asymmetry in use of semantic category information

128 DAVID HINES er al

half-field (right hemisphere), to the right visual half-field (left hemisphere) and at fixation (available to both hemispheres). Pictures of common objects were used as the visual stimuli, since such pictures can be recognized well by either hemisphere [ 111. Pictures of common objects also provide a larger priming effect than word stimuli on the semantic priming task,

when subjects are requested to name the target stimuli [l, 81. Thus, the experiment was designed to isolate the priming stimulus to a particular

hemisphere through subthreshold presentation and then to test for response facilitation in naming a second, target, stimulus when the target was presented initially to the same or opposite hemisphere.

EXPERIMENT 1

Subjrcrs. Twenty-one right-handed undergraduates participated as subjects in the experiment (15 females. six

males). Subjects who had uncorrected visual acuity deficits in one or both eyes were excluded. Subjects received extra

credit in an introductory psychology course in return for participating. Handedness was assessed through self- report.

Equipment and uisuaIstimuli. The stimuli were presented using a Gerbrands Harvard T4A Tachistoscope Reaction time was measured using a Gerbrands G1270 digital msec clock/counter and a Gerbrands Gl355 voice-operated relay. The timing was Initiated electronically by the presentation of the target stimuli and was stopped by the gating

of the voice-operated relay when the target was named. The prime and target pictures were taken from the Peabody Picture Vocabulary Test. Each picture was reduced to

fit within a 2.5 x 3.8 cm square (about 42”:; of the original size). Twenty drawings were selected to form IO pairs of semantically related (e.g. dog-cat. axe-saw) prime and target stimuli. Each prime was then randomly recombined with one of the other target pictures to form 10 unrelated pairs (e.g. dog-saw, axe-cat).

Each stimulus was mounted on a white 4 x 6 index card for presentation. Each prime and target picture was presented in three different positions: centered at fixation, centered 2.5 cm (2.7’ of visual arc) to the right of fixation and centered 2.5 cm to the left of fixation.

Prucedwr. Each subject was tested in two sessions separated by at least one day. In Session 1, the naming

threshold for each of the 10 prime stimuli was individually determined at each of the three presentation positions. In Session 2, the reaction times for the targets in the related and unrelated pairs were measured.

Tltrrsltold determinarion. The IO prime stimuli (cat, bat, banana. chair, gun, axe, bucket, truck, drum and tie) were presented to the left visual held. right visual field and at fixation. The order of presentation was random in regard to

both stimulus and field position, with the restriction that the same stimulus was never presented twice in succession. Subjects were asked to name each stimulus as it was presented and encouraged to guess if they were not entirely

sure of its identity. However, subjects were not forced to guess if they reported they had no idea of the stimulus. On each trial, a central fixation circle was displayed for 700 msec followed by the prime to the left eye, followed by

a cognitive mask (composed of fragments of pictures of objects not used as the priming or target stimuli) to the right eye for 50 msec. Each stimulus was initially presented for 80 msec. The exposure time for each stimulus at each

position was individually increased by 5 msec steps until a given prime was correctly named on four trials. The duration was then decreased 5 msec for each stimulus at a given position after a correct identification until the

stimulus was missed on four consecutive trials at that position. This duration was used for the particular prime at that position during Session 2. Thus, during Session 2. each prime was presented for a duration just below naming

threshold for the subject at that particular position. Derermi,rarion oj‘rracrion tintrs. At the beginning of Session 2. subjects named the 10 prime and target stimuli at

suprathreshold durations, Subjects were then administered 20 practice trials (in order to stabilize reaction time) followed by 180 scored trials. On each trial, subjects viewed a fixation card binocularly for 700 msec followed by a prime stimulus to the left eye at subthreshold duration as determined during Session 1. The prime stimulus was followed by the cognitive mask to the right eye for 50 msec, followed by the fixation card binocularly for 450 msec, and lastly the target stimulus binocularly for 150 msec. Subjects were instructed to fixate the center of a small circle each time the fixation card appeared. to report the second (target) stimulus as rapidly as possible and then to report the first (prime) stimulus. Any errors in order of report or target identification were corrected. Errors in target identification were extremely rare (less than 1%) on the experimental trials and were not scored separately.

Results

The mean reaction times for all conditions where both stimuli were directly available to the left hemisphere (presented at fixation or in the right visual field) are contrasted to those

Page 3: Hemispheric asymmetry in use of semantic category information

HEMISPHERIC ASYM.WTRY AND SE.MANTIC CATEGORIES 129

conditions where one or both stimuli went initially to the right hemisphere (presented in the

left visual field) in Table 1. The mean priming effect for the four conditions where both stimuli were directly available to the left hemisphere was highly significant, r(20) = 2.86, P < 0.005. No significant priming effect was found for the five conditions where one or both stimuli went initially to the left visual field-right hemisphere, t(20)=0.58, N.S. Consistent with these results, a significantly larger priming effect was found for the four conditions where both stimuli went directly to the left hemisphere than for the five conditions where at least one stimulus went to the left visual field-right hemisphere, t(20)=2.29, P<O.OX.

Table 1. Mean reaction times (in msec) for related and unrelated pairings as a function of prime and target visual field (VF) positions

(a) Both stimuli directly available to left hemisphere

Prime position Fixation Right VF Fixation Right VF Target position Fixation Fixation Right VF Right VF Mean

Unrelated pairs 707 707 734 722 718 Related pairs 692 687 693 683 689 Priming effect 15 20 41* 39’ 29’

(b) At least one stimulus presented to left visual field-right hemisphere

Prime position Left VF Fixation Right VF Left VF Left VF Target position Left VF Left VF Left VF Fixation Right VF Mean

Unrelated pairs 732 718 731 705 735 724 Related pairs 723 739 706 711 718 719 Priming effect 9 -21 25 -6 17 5

‘P<O.O5

As can also be observed in Table 1, the means and size of the priming effect seem to be consistently positive for the four conditions when both stimuli were available to the left hemisphere. However, the priming effects for the five conditions where one stimulus went to the right hemisphere are more variable with some values negative. Negative priming effects (i.e. higher values on related trials) are sometimes associated with physical similarity between stimuli [3]. Some of the semantically related stimuli in this study were physically similar (e.g. dog-cat, car-truck) although others were not (e.g. axe-saw, banana-pear). It is possible that the positive, semantically related effects were predominant when both stimuli went to the left hemisphere, but the negative, physical similarity effects sometimes predominated when both stimuli went to the right hemisphere. The latter suggestion remains speculative, however, since none of the negative effects were statistically significant.

Subjects were instructed to report the first (prime) stimulus in Session 2 after reporting the second (target) stimulus. Subjects reported a prime stimulus on 29% of all trials, with half of the primes reported a correct identification (i.e. 14.5% correct). The number of primes correctly identified was not significantly related to the magnitude of the subjects’ semantic priming effect, r(20) = 0.14. Recognition of primes was highest from fixation (19.4%) followed by the left visual field (14.2%) followed by the right visual field (10%); these differences were statistically reliable F(2, 80) = 9.01, P < 0.01.

Page 4: Hemispheric asymmetry in use of semantic category information

430 DAVID HINES er al.

Discussion

The decreased reaction time to targets in related pairs requires both the activation of semantic category information in long-term memory when the priming stimulus is processed and the use of this information to speed naming when the target stimulus is processed. The present data demonstrate that both of these processes are performed efficiently by the left hemisphere. A large priming effect (39 msec) occurred in the condition where both the first and second stimulus were presented to the right visual field-left hemisphere. In contrast, no

priming effect was found when either the prime or target was initially available only to the right hemisphere. No significant priming effect occurred in those conditions where either the prime or target stimulus was presented to the left visual field-right hemisphere.

There is a possible methodological artifact in Experiment 1 due to the monocular presentation of the subthreshold primes to the left eye and the masking stimulus to the right eye. When the prime was presented only to the left eye, stimuli to the right visual field were projected to the temporal retina and transmitted to the left hemisphere via the uncrossed optic nerve pathway. In contrast, stimuli to the left visual field projected to the nasal retina

and were transmitted to the right hemisphere via the crossed optic nerve pathway. Thus, the differences between stimuli in each visual field could have resulted from retinal or optic nerve differences rather than hemispheric differences. Monocular presentation is used in masking experiments to ensure that the masking takes place at the level of the brain rather than peripherally at the level of the retina [3]. However, subthreshold priming can also be reliably demonstrated with binocular presentation of the prime and mask. [l, 81.

In Experiment 2, binocular presentation of the prime stimuli and the mask was used, in order to control for possible retinal or optic pathway asymmetries. In addition, a modification of the threshold procedure developed by HISES, CZERWINSKI and SAWYER [5] wm used, in order to reduce recognition of the prime stimuli during the reaction time trials. The largest difference in priming effect in Experiment 1 was found in those conditions where the prime was presented at fixation and the target to either the left or right visual field. Therefore, these conditions were also used in Experiment 2.

Experiment 2 also has an additional control for any possible attentional bias favoring either visual half-field [7]. All primes were presented at fixation; thus, by the reaction time trials in Session 2 subjects should have been well trained to fixate the center position. The targets were randomly alternated to the left and right visual fields. In order to test for possible bias toward attending to targets in either visual field, baseline trials with only a target presented were randomly interspersed with the priming trials. If subjects were preferentially attending to either visual field, they should have been able to name the targets from that field more quickly in the target alone condition.

EXPERIMENT 2 M er hurl

Subjecrs. Nineteen additional right-handed subjects (14 females, five males) were selected as in Experiment 1. Equipment and visual stimuli. The equipment used was identical to Experiment 1. The visual materials were

prepared following the same procedure except that 24 pictures were selected from the Peabody Picture Vocabulary Test and from children’s coloring books. Of these, I2 were used as prime and 12 as target stimuli. The 12 prime stimuli were relatively symmetrical in order to provide similar information from the left and right halfofeach prime. As an additional control, a photographic reversal of each prime was used for half of the subjects. No difference of any kind was found related to use of the original or reversed primes and the data were combined across this variable for all analyses to be reported.

Procedure. As in Experiment 1, each subject was tested in two sessions, first to determine the prime threshold and

Page 5: Hemispheric asymmetry in use of semantic category information

HE.MISPHERIC ASYWMETRY AND SENANTIC CATEGORIES 131

then to measure the target reaction times. The reaction times were measured under three different conditions (related pairings. unrelated pairings and a target-only baseline condrtion) for both the right visual field and left visual field targets. The condition and visual field was randomly alternated within blocks of six trials. each of which contained one trial in each combination of conditions.

Thresholci session. The 12 pictures used as primes were divided into two groups of six physically distinct pictures in order to reduce subject’s confusion between pictures. The first group consisted of a bus, carrot, apple, glass, cake and drum. The second group consisted of a table, cat, ball. comb. jacket and TV. The procedure used in Experiment 1 was then followed, except that each prime stimulus was presented binocularly and was initially exposed for 90 msec before presentation of the mask. The masking stimulus was also presented binocularly and was exposed for 200 msec following the prime on each trial. For nine subjects, theexposure duration at which a parocular prime was missed on four consecutive trials was used as the duration for that prime during the reaction time trials (as in Experiment I). For the remaining 10 subjects, this duration was reduced by loo,; before being used as the duration for the prime in the reaction time trials. Reducing the threshold by lop/, reduced recognition of primes on the reaction time trials (from 3 to 0%) but did not significantly alter any of the reaction times to the target stimuli. The data are therefore combined across this variable for the analysis to be reported.

Determination ofreacrion times. At the beginning of the second session, subjects named the 12 primes and 12 targets at suprathreshold durations. Subjects were then administered 24 practice trials, followed by 72 scored trials. On each trial a fixation stimulus was presented for 700 msec. followed by a prime stimulus, followed by the pattern mask for 200 msec. followed by the fixation stimulus for 300 msec. followed by the target picture for 450 msec. All stimuli were presented binocularly.

Results

The mean reaction times for each condition are shown in Table 2. The results replicate those of Experiment 1, with a reduced reaction time for the related stimulus pairs compared to the unrelated pairs only when both stimuli were directly available to the left hemisphere.

Table 2. Mean reaction times (in msec) as a function of stimulus pairings and target visual field

Target Target visual field only

Stimulus pairings Unrelated Related

pairs pairs Priming

effect

Left 611 620 630 - 10 Right 615 638 623 15’

‘P<O.O5.

No differences were found between the left and right visual field target reaction times when only the targets were presented in the baseline trials, F (1,lS) < 1. The data for the related and unrelated pairings were then analyzed by analysis of variance, with type ofpairing and target visual field as within group variables. Neither main effect was significant, with each F < 1; however, the interaction between type of pairing and target visual field was significant, F (1, 18)= 8.88, P~0.01. Subsequent planned comparisons showed that the right visual field targets were recognized significantly faster in the related pairings than the unrelated pairings. However, for the left visual field targets the higher reaction times for the related pairings were not statistically significant.

In comparison to Experiment 1, subjects recognized relatively few of the prime stimuli during the reaction time trials (1.5% correct). Only two of the 19 subjects recognized any primes correctly during the reaction time trials.

The reduction in the number of primes recognized also produced a much greater consistency in the reaction times for each subject. When subjects recognized the very briefly

Page 6: Hemispheric asymmetry in use of semantic category information

presented primes on the reaction time trials in Experiment 1, it often delayed their response to the target stimulus. These increased latencies would not have changed the size of the priming effect, since they should have occurred equally often on related and unrelated trials, However, the very long response times on occasional trials did increase the size of the error variance. (See [S] for a similar finding.) With the reduced recognition of primes in Experiment 2, the size of the differences between conditions needed for statistical significance was comparable to that reported in other studies of subthreshold priming [l, 81.

The reduced recognition of primes might also have contributed to the smaller priming effect for the right visual field targets found in Experiment 2 as compared to the comparable condition in Experiment 1. However, there were also differences in the set of pictures used and in the use of binocular rather than monocular presentation of the prime and target, which might equally have altered the size of the effect.

GENERAL DISCUSSION

The experiments reported avoid the major interpretative ambiguity associated with studies which relate visual field asymmetries to hemispheric function, because they avoid direct comparisons of recognition or report between the left and right visual fields. Such direct comparisons can be ambiguous because visual field asymmetry is subject to the influence of attentional and order of report effects, as well as reflecting the processing capacities of the two hemispheres [7,9]. However, the present experiment compared reaction times for the same target stimulus presented to the same visual field. The only difference in procedure was the semantic relatedness of the prime and target pairings. Since the related and unrelated pairings were randomly alternated across trials, it was not possible for attentional or order ofprocessing bias to alter differentially target reaction time to the related and unrelated pairings. These considerations indicate that the differences in the semantic priming effect associated with visual field position found in the present study can be attributed to differences in information processing between the left and right hemispheres.

The results of both Experiments 1 and 2 indicate that semantic priming is a task which requires the processing capacity of the left hemisphere. No hemisphere differences were found either in recognition of the primes (in Experiment 1) or in the response times to name the targets presented alone (in Experiment 2). However, a semantic priming effect was found in both Experiments 1 and 2 when both the prime and target were initially presented to the left hemisphere but not when the prime or target were initially presented to the right hemisphere.

The final question to be considered is why the right hemisphere fails to show any semantic priming effect. The right hemisphere has been found to produce a larger affective discrimination response than the left hemisphere when nonverbal shapes are masked below recognition threshold [lo]. Thus, the right hemisphere is not inferior in the initial processing of pattern masked nonverbal stimuli. This suggests that the right hemisphere does not use the semantic relationship between the prime stimulus and target stimulus when naming the target object, either because such information is unavailable to the right hemisphere or because the right hemisphere is specialized for using other types of information (such as the visual characteristics of the stimulus).

Ackno\~ledgemenrs-These experiments were supported by research grants from Ball State University. The authors are grateful to Julie Clista. Lisa Warren and Catherine Zmachinski for their assistance in collecting data and to Glenn Davidson, Robin Barr and Darrell Butler for their helpful comments on the manuscript.

Page 7: Hemispheric asymmetry in use of semantic category information

HE.MISPHERIC ~sywmmy ASD SE.MANTIC CATEGORW 333

REFERENCES

I. CARR, T. H., MCCALLEY, C., SPERBER, R. D. and PARS(ELEE, C. M. Words. pictures and priming: On semantic activation, conscious identification. and the automaticy of information processing. JI. esp. Psycho/.: Humon Percept.: Perform. 8, 157-777, 1982.

2. COLLINS, A. M. and LOFTUS. E. F. A spreading-activation theory of semantic processins. Psychol. Rec. 82, 407-428. 1975.

3. FOWLER, C. A., WOLFORD. G., SLADE. R. and TASSINARY, L. Lexical access with and without awareness. JI. Exp. Psychol.: Cert. 110, 341-362, 1981.

4. HINES, D. Visual information processing in the left and right hemispheres. ~Veurops~cholo~ia 16,593~600, 1978. 5. HINES, D., CZERWINSEI. M. and SAWYER, P. K. The effect of conscious identification of a prime on naming and

categorizing picture and word targets. Manuscript in preparation. 6. HINES, D., SATZ. P. and CLE.MESTISO. T. Perceptual and memory components of the superior recall of letters

from the right visual half-fields. .Veuropsychologia 11, 175-180. 1973. 7. KINSBOURNE. M. The mechanism of hemispheric control of the lateral gradient of attention. In r(rtenrion and

Perlormonce Y, P. M. A. RABBIT and S. DORNIC (Editors). Academic Press, new York. 1975. 8. MCCAULEY, C.. PARXELEE. C.. SPERBER, R. and CARR, T. Early extraction of meaning from pictures and its

relation to conscious identification. JI. esp. Psvchol.: Human Percept. Perform. 6, 265-276, 1980. 9. MOSCOVITCH. M. and KLEIN, D. Mateiial-specific perceptual interferknce for visual words and faces:

Implications for models of capacity limitations, attention. and laterality. Jl. e.rp. PsJchol.: Human Prrcepr. Perfirm. 3, 59&604. 1980.

10. SEAMON. J. G., BRODY. N. and KAUFF. D. M. Affective discrimination ofstimuli that are not recognized: Effects of shadowing, masking, and cerebral laterality. JI. exp. Psychol.: Learn. Mem. Cog. 9, N-555, 1983.

I I. YOUNG, A. W.. BIOX, P. J. and ELLIS. A. W. Studies toward a model of laterality effects for picture and word naming. Brain Lang. 11, 54-65, 1980.