- Email: [email protected]
Visual field differences in an object decision task
BRAIN AND COGNITION 19,
195-207 (1992)
Visual Field Differences in an Object Decision Task MELANIE Polytechnic
of East London,
VITKOVITCH
The Green, London,
El5 4L2 United Kingdom
AND GEOFFREY UNDERWOOD University of Nottingham,
Nottingham,
NG7 2RD United Kingdom
Two experiments are reported which investigate hemispheric processing in an object decision task. Experiment 1 used 40 pictures of objects, and 40 pictures of nonobjects, and subjects decided manually whether each lateralized stimulus represented an object. Results indicated an interaction between visual field and response (yes versus no). There was a right visual field advantage for positive responses, but no difference between visual field for negative responses. Positive responses were faster than negative responses, and this effect was more marked for right visual field presentations. These results were replicated in a second experiment. The results are interpreted as reflecting a left hemisphere superiority at accessing stored structural descriptions of known objects. The possibility that left and right hemispheres use different methods of carrying out the task is also discussed. 8 1992 Academic Press. Inc.
INTRODUCTION In the divided visual field literature, there is a relative lack of research on picture recognition, as compared to the numerous studies on word and face recognition. The present study examines hemispheric differences with respect to one particular aspect of object recognition. Young and Ellis (1985) have pointed out the need to relate data from studies of cerebral asymmetry to detailed theoretical models, and currently, there are a number of models of object recognition available (Humphreys, Riddoch, & Quinlan, 1988; Seymour, 1979; Snodgrass, 1984; Warren & Morton, 1982). These models generally suggest that three representational Reprint requests should be addressed to Melanie Vitkovitch, Department of Psychology, Polytechnic of East London, The Green, Romford Road, Stratford, London, El5 4L2 United Kingdom. 195 0278-2626192$5.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.
196
VITKOVITCH
AND UNDERWOOD
stages are involved in object recognition. Visual analysis of an object leads to a description of the physical features, and this input must then be matched against a stored structural description, so that the object may be recognized as a known object. Semantic information may then be accessed, and, as a third and final stage, the name of the object may be retrieved. It is not always possible to relate the results of visual field studies specifically to these models. An early conceptualization of the specialization of the two hemispheres was simply in terms of stimuli type, with the left hemisphere (LH) specialized for verbal stimuli and the right hemisphere (RH) specialized to deal with nonverbal stimuli. A number of studies (e.g., Bryden & Rainey, 1963; Dee & Fontenot, 1973) were designed around this conceptualization and compared the processing of nonverbal stimuli (such as geometric or nonsense shapes) with verbal stimuli (e.g., pictures of objects) in tasks such as shape matching. It was generally concluded that the ease with which a verbal label could be attached to a shape determines the visual field asymmetry, with easily labeled shapes, such as objects, leading to a right visual field advantage (RVFA), and geometric or nonsense shapes occasionally demonstrating a left visual field advantage (LVFA). Davidoff (1982) and Young and Ratcliff (1983) also state that the extent to which a form is named may be important. However, from these studies it is not easy to specify at which stage in the processing of objects any RVFA may be emerging. A task such as matching need not involve any stored representation stages. A further characterization of the two hemispheres was in terms of preferred modes of processing, rather than types of stimuli, and several dichotomies have been proposed, e.g., LH as an analytical processor and RH as a holistic processor. Some studies (e.g., Magnani, Mazzucchi, & Parmon, 1984; Tomlinson-Keasey & Kelley, 1979) have used the same pictures or shapes with different tasks that encourage one mode of processing or the other, and there is some support for an account which suggests that it is the processing demands of a task, rather than the nature of the stimuli, which may be a major determinant of cerebral asymmetry. In a similar vein, it has been suggested (e.g., Klatzky, 1972) that when picture stimuli are matched on a nominal basis, as opposed to a physical basis, then LH advantages should be apparent. The implication of this claim is that it is the process of attaching a label to the picture which leads to LH advantages, not the easily labeled picture per se. This latter idea might suggest that it may only be the final stage of object recognition (name retrieval) which leads to the LH advantage. In general, it is the case that the naming of lateralized picture stimuli leads to an LH advantage, possibly because of LH involvement in output processes (see Underwood & Whitfield, 1985), although this is not always the case (Young, Bion, & Ellis, 1980). It is generally accepted that the
LATERALIZED
OBJECT DECISION
197
RH is superior at the early stages of visuospatial processing (see Davidoff, 1982, and Young & Ratcliff, 1983, for reviews). The stages of recognition which, according to the models above, are thought to occur between visual processing of the stimuli, and their naming, have received much less attention. For example, there have been no studies of the normal population that, to our knowledge, have directly addressed the issue of hemispheric involvement in the earliest stage of object recognition (i.e., matching the visual input to a stored structural description). This contrasts with a recently accumulating body of research concerning RH involvement in the early stagesof face processing (e.g., Hay, 1981; Newcombe, De Haan, Ross, & Young, 1989; Rhodes, 1985; Young, Hay, & McWeeny, 1985; Young, Hay, McWeeny, Ellis, & Barry, 1985). In the clinical literature, there is one study which would suggest that the early stage of object recognition is lateralized to the RH (Warrington & Taylor, 1978). Young, Bion, and Ellis (1980), and Young and Bion (1981) do suggest, from their studies of picture naming in the normal population, that both hemispheres may be involved in the identification of objects, although they stress the need for further research to clarify the level at which RH identification of pictures is achieved. In the following two experiments, we use an object decision task in order to address the issue. The object decision task was first introduced by Kroll and Potter (1984) and bears some similarity to the lexical decision task (though see Kolers & Brison, 1984) and to face decision tasks. In the object decision task, nonobjects, constructed from combining parts of different true objects, are presented randomly with true objects. Subjects are required to decide whether the stimulus represents a true object or not. Logically, deciding whether a line drawing depicts a known object or not involves access to stored structural descriptions. Riddoch and Humphreys (1987), who use the task in a clinical context, maintain this point. Lupker (1988) also suggeststhat the task should tap the entry point to semantics, which from the model of object recognition outlined above is shown to be at the structural description stage. Lupker failed to find any semantic priming effects using the object decision task, suggesting that the task does not automatically involve semantic processing. Kroll and Potter argue that the objects are not being named during the task, and clearly no overt naming is required (avoiding the problem of LH speech output processes). There would seem to be some agreement that the task taps the early stage of processing, and for this reason, we investigate whether this task leads to lateral differences in the normal population. It is difficult from the preceding review of studies to make any firm predictions, although the RH superiority at visuospatial processing, and the studies by Warrington and Taylor, Young and Bion, and Young et al. suggest that the RH should not be at any particular disadvantage.
198
VITKOVITCH
AND
UNDERWOOD
EXPERIMENT 1 Subjects
Sixteen subjects took part in this experiment. All were undergraduates or postgraduates at the University of Nottingham, were right-handed by self-report, and used their right hand for writing in the normal position. All had normal or corrected-to-normal eyesight. Stimuli
Pictures of objects were selected from the standardized set of Snodgrass and Vanderwart (1980). For reasons related to other research in the same laboratory, 20 pictures of fruit and vegetables, and 20 pictures of manmade objects were used. In addition to the 40 positive stimuli, 40 pictures of nonobjects were selected from the set constructed by Kroll and Potter (1984). These nonobjects were formed by tracing parts of pictures of true objects and then regularizing the stimuli. The set of 40 nonobjects were carefully selected so that the true objects and nonobjects did not differ in any obvious way with respect to superficial characteristics. The stimuli were photographed (the nonobjects and objects were photographed to the same size) and mounted on slides. Two slides were prepared of each stimulus-one in each visual field. These slides were mirror images of each other. Apparatus and Display
Slides of pictures were displayed on a blank white screen by use of a Kodak Carousel back-projector. The shutter rise time was 1.5 msec. The projector was controlled by a Rockwell Aim 65 microcomputer, mediated by an interface. The illuminated field of the screen measured 9” by 5”. The illuminated level of the preexposure field was 45 candela/m2. The illuminated level of the exposure field was 286 candela/m2. A permanent fixation dot was centered on the illuminated screen. A chin rest, fixated to a table, was positioned in front of the back-projection screen such that, when the subject was seated, the distance from the back-projection screen was 42”. The visual angle from fixation point to the right side of the picture in the LVF and the left side of the picture in the RVF was 2”2’43”. The pictures were measured for vertical and horizontal angles subtended. The mean vertical angle was 2”2’43”, and horizontal angle was 2”34’40”. A small box containing the two response keys was positioned centrally in front of the chin rest and connected to the computer. The display procedure was as follows. A trial started with a warning tone. During this time the preexposure field was displayed. One second after the tone, the slide was displayed for an interval of 150 msec. With the termination of this display, the preexposure field reappeared. The subject made his/her response. There was then an interval of 1 set before
LATERALIZED
OBJECT DECISION
199
the next trial started, which would not commence until a response was made. Design and Procedure
The task required subjects to respond positively or negatively according to whether the displayed stimulus represented a real object or not. Stimuli were presented in either the LVF or RVF. There were therefore two within-subject factors-response type and field of presentation, each with two levels. The stimuli were presented in three blocks, with a short break in between (there were 54 experimental stimuli in two blocks, and 52 in a third block). The first eight trials in each block were practice stimuli. There were an equal number of positive and negative stimuli in each block, and an equal number of LVF and RVF stimuli. Apart from this, the stimuli were randomly allocated to blocks. Slides within each block were rerandomized every four subjects. The order of blocks was random. Subjects were instructed to fixate on a central dot at all times. They were told to press a “yes” key if the stimulus presented depicted a true object and to press the “no” key if the stimulus did not represent a true object. It was decided randomly whether subjects should use their index finger of the right hand for the positive key, and their left hand for the negative key, or vice versa. In addition to the practice stimuli within each block, subjects were given a practice session at the beginning of the session. They were instructed to respond as quickly and as accurately as possible. Results
The error rate in this experiment was found to be generally quite high, with a few subjects demonstrating very high error rates. A criterion of 85% correct responses (overall) was set, and any subjects with over 15% errors were excluded from the analysis. Five subjects were excluded from the analysis. Mean reaction times (RTs) were calculated for each subject for each condition, and the mean of these means for each condition is shown in Table 1. An analysis of variance indicated a main effect of response (F(1, 10) = 32.2, p < .OOl; MS, = 3065.17), with positive responses faster than negative responses, and an interaction between visual field and response (F(1, 10) = 9.48, p < .05; MS, = 699.42). There was no main effect of visual field (F < 1; MS, = 1290.46) The interaction was further analyzed using simple main effects (Kirk, 1968). For the positive responses, there was a reliable difference between LVF and RVF (F(1, 20) = 4.74, p < .05), with faster responses to RVF stimuli. There was no difference between LVF and RVF stimuli for the negative responses (F(1, 20) = 2.17,
200
MEAN
VITKOVITCH
AND
UNDERWOOD
TABLE 1 RTs (msec) AND PERCENTAGE ERRORS FOR THE POSITIVE AND NEGATIVE THE LATERALIZED
OBJECT DECISION
TASK IN EXPERIMENT
Positive
X RTs Error
RESPONSES IN
1
Negative
LVF
RVF
LVF
RVF
633 15%
604 13%
704 6%
723 11%
p > .05). The difference between positive and negative responses was significant for both the RVF (F(1, 20) = 41.57, p < 0.001) and LVF (F(1, 20) = 14.39, p =C.Ol), although inspection of the means shows a
much stronger effect in the RVF than the LVF. There was a positive skew to the RT data, and, accordingly, the data were submitted to a logarithmic transformation (Kirk, 1968; Howell, 1987). Analysis of the transformed data gave the same results, with respect to main effects and interaction and with respect to analysis of simple main effects. Errors were also analyzed. The main effect of response (more incorrect positive responses) and the interaction between response and VF just failed to reach significance (F(1, 10) = 4.90, p < .lO; MS, = 58.24) and (F(1, 10) = 4.72, < .lO; MS, = 30.81), respectively). The main effect of visual field was not reliable (F(1, 10) = 3.13, p > .lO; MS, = 9.41). The marginally reliable interaction reflects a reduced error rate for the LVF negative stimuli. Discussion
The analysis of positive and negative RT responses indicated a significant interaction between VF and response. Analysis of simple main effects revealed an RVFA for positive responses, but no significant VF difference for negative responses, although mean RTs were faster for LVF negative responses than RVF negative responses. Positive responses were overall quicker than negative responses, although this effect was more marked for RVF stimuli. The interaction between VF and response argues against any explanation for VF arising from an attentional bias in favor of one VF or the other, as suggested by theories such as Kinsbourne’s (1970). It is also the case that explanations in terms of lack of fixation, or scanning tendencies, are less relevant. These kinds of explanations would favor an overall VF advantage, unqualified by interactions with stimuli type (see Patterson & Besner, 1984; Young & Ellis, 1985; Zaidel & Schweiger, 1984). However, an interaction between VF and response type could possibly reflect a response bias for one hemisphere. For example, the quicker responses
LATERALIZED
OBJECT
DECISION
201
for positive stimuli in the RVF could reflect a tendency for the LH to respond “yes” readily. Similarly, the trend for faster negative response in the LVF might reflect an RH readiness to respond “no.” If this were the case, then error rates could be expected to reflect these kinds of bias. The analysis of errors indicated an interaction (marginally reliable) between VF and response. For the RVF responses, there was no difference in error rate for positive and negative responses, which is inconsistent with the suggestion that the faster RVF positive RTs were due to a response bias in favor of “yes” responses. If this were the case, error rates should be raised for “no” responses. For the LVF responses, error rates between positive and negative responses differed somewhat, with a reduced error rate for “no” responses. Furthermore, LVF negative responses were more accurate than RVF negative responses. This could possibly reflect a bias to respond “no” to LVF stimuli. By this account, one might expect more errors for positive LVF stimuli than positive RVF stimuli, and there is a slight increase in LVF error rates for positive stimuli, although it is not particularly marked. Given the possibility that a response bias cannot be completely ruled out from this error data, a replication study is reported which examines the reliability of the results. EXPERIMENT
2
This experiment is similar to Experiment 1, although only a subset of the stimuli were used, thereby shortening the session. The 20 fruit and vegetable stimuli only were used as objects, and accordingly, 20 nonobjects were chosen from the larger set. The nonobjects were again selected carefully so that any discrimination between this subset of objects and nonobjects could not be readily achieved by figural differences between the two stimuli sets. For example, the nonobjects had rounded or elongated shapes and their outlines were generally nonangular, to match the shapes of fruit and vegetables (e.g., apple, carrot). There was only one block of 80 stimuli, and these were preceded by a practice session. The apparatus and the procedure were the same as for Experiment 1. Twelve new subjects were tested, from the same population as the first experiment. Results The error rate in this experiment was less than in Experiment 1, although two subjects did not reach the criterion of 85% correct. These subjects were excluded from the analysis. The mean RT for each subject for each condition was calculated, and Table 2 presents the mean of these means. The data were analyzed as in Experiment 1. The main effect of response (faster positive responses) was marginally reliable (F(1, 9) = 3.94, p > .05; MS, = 18758.83). The main effect of VF was not reliable (F > 1;
202
VITKOVITCH
AND TABLE
MEAN
UNDERWOOD 2
RTs (msec)
AND PERCENTAGE ERRORS FOR THE POSITIVE AND NEGATIVE THE LATERALIZED OBJECT DECISION TASK IN EXPERIMENT 2 Positive
fi RTs Error
RESWNSES IN
Negative
LVF
RVF
LVF
RVF
685 7%
660 5%
754 9%
763 10%
MS, = 1221.12), but the interaction between VF and response was significant (F(1, 9) = 8.81, p < .05; MS, = 331.82). The interaction was analyzed using simple main effects. Positive responses were significantly faster than negative responses for RVF stimuli, but the effect of response for LVF stimuli was unreliable (F(1, 18) = 2.49, p > .05). The difference between RVF and LVF for the positive responses (faster positive responses) just failed to reach significance at the 5% level (F(1, 18) = 4.12, p > .05). For the negative responses, there was no difference between VF (F < 1). As in Experiment 1, the distribution of the data demonstrated a positive skew, and logarithmic transformation of the data and subsequent analysis gave the same pattern of effects as the raw data, with effects now reaching the 5% significance level. The main effect of response was significant (F(1, 9) = 5.88, p < .05), as was the interaction between VF and response (F(l) 9) = 9.45, p < .Ol). The main effect of VF was not significant (F < 1). The simple main effect of VF for positive responses was significant (F(1,18) = 5.49, p < .05), but the effect of VF for negative responses was not reliable (F < 1). The simple effect of response type for RVF was significant (F(1, 18) = 8.74, p < .Ol), and the effect of response type for LVF was marginally reliable (F(1, 18) = 3.28). As in Experiment 1, the effect of response was more pronounced for RVF stimuli as compared to LVF stimuli. Errors were also analyzed, but there were no effects (main effect of response (F(1, 9) = 2.31, p > .05; MS, = 2.12): main effect of VF (F < 1; MS, = .43) and the Response x VF interaction (F(1, 9) = 1.53, p > .05; MS, = 1.04)). Discussion The interaction between VF and response was replicated in this second experiment. The pattern of means in both experiments is the same, with RVF positive responses faster than LVF positive responses. There was no difference between LVF and RVF negative responses, although again, as in Experiment 1, LVF responses were slightly faster. The difference between positive and negative responses was significant for RVF responses. LVF positive responses do show faster response times than LVF
LATERALIZED
OBJECT DECISION
203
negative responses, though in both experiments, this effect is not as marked as for RVF responses. The error analysis in this second experiment, however, did not show any significant differences between conditions. This suggeststhat response bias does not account for the interaction between VF and response type in the RT analysis. It can be noted that error rates in this second study are less than in the first experiment. This could be due to the shorter experimental session, although, given that the reduction in errors is shown mainly for the positive stimuli, it is more likely that the number of errors in Experiment 1 were due to particular stimuli not present in Experiment 1. In fact, inspection of the errors does indicate that subjects tended to consistently make errors to certain stimuli, suggesting these stimuli were not easily recognizable in the context of the present task. GENERAL DISCUSSION In two experiments, we examined visual field differences in an object decision task. Subjects were required to judge whether lateralized stimuli represented true objects or not. There was some difficulty in performing this task, evident by relatively high error rates. At least part of this difficulty was due to certain objects which, in the context of the nonobject fillers, were apparently not very easy to recognize. Nevertheless, an interesting pattern of results emerged, and this was consistent over both experiments. Analysis of the RTs to lateralized true objects and nonobjects indicated both a main effect of response and an interaction between visual field and response type (“yes” versus “no”). The pattern of means demonstrates that positive responses to true objects were faster than negative responses to nonobjects. This result is consistent with data from object decision tasks performed under central viewing (Kroll & Potter, 1984; Lupker, 1988). It suggests generally that the task is not being performed on any superficial physical characteristics which would allow an easy discrimination between true objects and nonobjects, without accessing stored structural descriptions. Were this the case, nonobjects could be rejected as quickly as true objects are accepted. However, the RT data for the LVF presentation of stimuli did show a less marked difference between true objects and nonobjects than the RVF data. The interaction between visual field and response type may be an indication that the LH and RH carry out the task in different ways, In the Introduction, it was noted that there have been suggestions that the LH analyzes stimuli in an analytical manner, the RH in an holistic fashion. The more pronounced difference between positive and negative responses for RVF stimuli may reflect, for example, an analytical feature comparison process adopted by the LH. If the LH analyzes the stimulus in terms of features, then partial activation may occur in stored structural descriptions
204
VITKOVITCH
AND UNDERWOOD
which contain those features. This would occur for both true objects, and nonobjects, given that the latter are constructed from the parts of true objects (see under Stimuli). Top-down processing could speed the recognition of true objects, but a time-consuming checking process may have to be implemented in order to reject the nonobjects. In contrast, the RH may analyze the stimuli in a more holistic fashion. Negative responses, since they do not as a whole match any true object, may be readily rejected, and hence positive and negative RTs do not differ so markedly for LVF stimuli. These suggestions could be tested by varying the degree of featural similarity between nonobjects and objects. Increasing the similarity between objects and nonobjects would then be expected to influence the responses to RVF stimuli, without affecting LVF responses. In both experiments, an RVFA was found for positive responses. There was also a slight tendency for faster responses to LVF nonobject stimuli (consistent with the studies mentioned in the Introduction which suggest LVFA for nonsense stimuli). However, the VF differences for negative responses were not significant. An absence of VF effects for negative responses is not uncommon in VF studies (Young, Hay, & McWeeny, 1985), and some dismiss this as inconsequential, providing that the positive responses demonstrate consistent advantages. However, Lambert (1982) provides a fuller discussion of the possible interpretations of VF advantages, which we consider here in the context of our own results. We have already argued that explanations such as attentional bias and scanning tendencies are unlikely in the present case (see Discussion to Experiment 1). With respect to hemispheric differences, an RVFA may be an indication that the processing of stimuli and decisions relevant to the task may be carried out by both hemispheres, with the LH simply more efficient at so doing (whether this is a result of similar methods used by both hemispheres, or different methods, as suggested above). In terms of the theoretical model of object recognition outlined, our data from the object decision task would suggest that the LH is better at either accessing stored structural descriptions, and/or at deciding a given stimulus corresponds to a relevant stored description. (Given the accepted RH superiority at visuospatial processing generally, it would seem unlikely that the LH superiority arises from an earlier visual analysis stage, or that it is due to the construction of an object description.) Alternatively, the RVFA may indicate that only the LH can carry out the processing relevant to the object decision task. The RVFA may then reflect the fact that RVF stimuli have more direct accessto the LH areas of the brain which carry out the processing. That is, the RVFA may be an indication that information from LVF stimuli may have to be transmitted to the LH (via the corpus callosum) for processing. In the present case, this would suggest the possibility that only the LH possessesstructural descriptions for known objects. However, the absence of VF asym-
LATERALIZED
OBJECT DECISION
205
metry for negative responses would then imply that the RH was capable of carrying out negative responses to nonobjects. As Lambert (1982) points out, this does not seem plausible. A further theoretical possibility considered by Lambert is that both LVF objects and LVF nonobjects may be transmitted via the corpus callosum to the LH. The interaction of VF with response would then imply that the LH processing of objects and nonobjects is differentially affected according to whether the input route is direct or indirect. Thus, the LH can process nonobjects equally rapidly, whether the stimulus is received directly via the RVF or indirectly from the LVF. True objects are processed more quickly via the direct route. Again, this explanation also seems unlikely. For the present, since we cannot determine the precise explanation for the RVFA for recognizing true objects, we conclude generally that the LH is superior at recognizing objects, whether it has a more efficient method than the RH or it is as a result of LH specialization. The results of the present study contrast with lateralized studies of face recognition which have used a face decision task. For example, in the study by Young, Hay, and McWeeny (1985), subjects had to decide if a presented stimulus depicted a true face or not. Stimuli were either line drawings of faces, or they were either moderately or highly scrambled faces. Young et al. found an LVFA when the scrambled faces resembled more closely true faces, and they concluded that the RH may be more involved in the construction of a detailed representation of a face. This would seem to support the conclusion drawn above that it is more likely that the LH advantage found in the object decision task study is due to processes which occur after the construction of the object description. A second possibility is that the LVFA found for the face decision task may be due to processing which may be unique, or more specific to faces. It has been suggested that the processing of faces may require configurational processing of the parts of the faces (Rhodes, 1985) and that this may favor RH processing. It can be noted that the contrasting set of results to the similar lateralized face and object decision tasks add strength to our argument that the present results are unlikely to be due to response bias, but genuinely reflect differences in hemispheric processing. It is of particular interest that the hemispheric processing of objects and faces would appear to differ at the early stages of recognition. This may be due to the nature of the processing required for the two different types of stimuli. A single experiment comparing face and object decisions, varying the nature of the nonobjects and nonfaces would be informative. In conclusion, the present data suggest an LH superiority at recognizing true objects in an object decision task. We suggest that this advantage is due to processesinvolving accessto, or comparison with, stored structural descriptions of objects. LH and RH processing of positive and negative
VITKOVITCH
206
AND UNDERWOOD
responses are shown to differ, and this could be an indication of different methods of performing the task. Further research on object decisions could follow the work on lateralized face decisions and examine the effect on visual field differences of varying systematically the degree of overall likeness between nonobjects and true objects. APPENDIX Positive stimuli used in Experiments 1 and 2: Ashtray, axe, barrel, balloon, bell, bowl, button, comb, cup, doorknob, drum, glass, guitar, pen, rolling-pin, shoe, thimble, top, wheel, whistle, apple, artichoke, asparagus, banana, carrot, celery, cherry, corn, lemon, mushroom, onion, orange, peach, pear, pepper, pineapple, potato, pumpkin, strawberry, tomato. (Only the last 20 fruit and vegetables were used in Experiment 2).
REFERENCES Bryden, M. P., & Rainey, C. A. 1963. Left-right differences in tachistoscopic recognition. Journal of Experimental
Psychology,
66, 568-571.
Davidoff, J. 1982. Studies with non-verbal stimuli. In J. G. Beaumont (Ed.), Divided visual field studies of cerebral organisation. London: Academic Press. Dee, H. L., & Fontenot, D. J. 1973. Cerebral dominance and lateral differences in perception and memory. Neuropsychologia, 11, 167-174. Hay, D. C. 1981. Asymmetries in face processing: Evidence for a right hemisphere perceptual advantage. Quarterly Journal of Experimental Psychology, 33A, 267-274. Howell, D. C. 1987. Statistical methodr for psychology. Boston: PWS-Kent. Humphreys, G. W., Riddoch, M. J., & Quinlan, P. 1988. Cascade processes in picture identification. Cognitive Neuropsychology, 5, 56-103. Kinsboume, M. 3970. The cerebral basis of lateral asymmetries in attention, Acta Psychologica, 33, 193-201. Kirk, R. E. 1968. Experimental design: Procedures for the behavioural sciences. California: Brooks/Cole. Klatzky, R. L. 1972. Visual and verbal coding of laterally presented pictures. Journal of Experimental
Psychology,
%, 439-448.
Kolers, P. A., & Brison, S. J. 1984. Commentary: On pictures, words and their mental representations. Journal of Verbal Learning and Verbal Behaviour, 23, 105-113. Kroll, J. F., & Potter, M. C. 1984. Recognizing words, pictures and concepts: A comparison of lexical, object and reality decisions. Journal of Verbal Learning and Verbal behaviour, 23, 39-66. Lambert, A. J. 1982. Right hemisphere language ability. 2. Evidence from normal subjects. Current Psychological
Reviews, 2, 139-152.
Lupker, S. J. 1988. Picture naming: An investigation of the nature of categorical priming. Journal of Experimental
Psychology:
Learning,
Memory
and Cognition,
14, 444-455.
Magnani, G., Mazzucchi, A., & Parmon, M. 1984. Interhemispheric differences in same versus different judgments upon presentation of complex visual stimuli. Neuropsychologia,
22, 527-530.
Newcombe, F., De Haan, E. H. F., Ross, J., & Young, A. W. 1989. Face processing, laterality and contrast sensitivity. Neuropsychologia, 27, 523-538.
LATERALIZED
OBJECT DECISION
207
Patterson, K., & Besner, D. 1984. Is the right hemisphere literate? Cognitive Neuropsychology, 1, 315-341. Rhodes, G. 1985. Lateraked processes in face recognition. British Journal of Psychology, 76, 249-271. Riddoch, M. J., & Humphreys, G. W. 1987. Visual object processing in optic aphasia: A case of semantic access agnosia. Cognitive Neuropsychologia, 4, 131-185. Seymour, P. H. K. 1979. Human visual cognition. London: Collier Macmillan. Snodgrass, J. G. 1984. Concepts and their surface representations. .lourna[ of Verbal Learning and Verbal Behaviour,
23, 3-22.
Snodgrass, J. G., & Vanderwart, M. A. 1980. Standardized set of 260 pictures: norms for name agreement,image agreement, familiarity and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174-215. Tomlinson-Keasey, C., & Kelley, R. R. 1979. A task analysis of hemispheric functions. Neuropsychologia, 17, 345-351. Underwood, G., & Whitfield, A. 1985. Right hemisphere interactions in picture-word processing. Brain and Cognition, 4, 273-286. Warren, C. & Morton, .I. 1982.The effects of priming on picture recognition. British Journal of Psychology, 73, 117-129. Warrington, E. K., & Taylor, A. M. 1978. Two categorical stages of object recognition. Perception,
7, 696-705.
Young, A. W., & Bion, P. J. 1981. Identification and storage of line drawings presented to the left and right cerebral hemispheres of adults and children. Cortex, 17, 459-463. Young, A. W., Bion, P. J., & Ellis, A. W. 1980. Studies toward a model of laterality effects for picture and word naming. Brain and Language, 11, 54-65. Young, A. W. & Ellis, A. W. 1985. Different methods of lexical accessfor words presented in the left and right visual hemifields. Brain and Language, 24, 326-358. Young, A. W., Hay, D. C., & McWeeny, K. H. 1985. Right cerebral hemisphere superiority for constructing facial representations. Neuropsychologia, 23, 195-202. Young, A. W., Hay, D. C., McWeeny, K. H., Ellis, A. W., & Barry, C. 1985. Familiarity decisions for faces presented to the left and right cerebral hemispheres. Brain and Cognition, 4, 439-450. Young, A. W., & Ratcliff, G. 1983. Visuospatial abilities of the right hemisphere. In A. W. Young (Ed.), Functions of the right cerebral hemisphere, London: Academic Press.
Zaidel, E., & Schweiger, A. 1984. On wrong hypotheses about the right hemisphere: Commentary on K. Patterson & D. Besner “Is the right hemisphere literate?” Cognitive Neuropsychology,
1, 351-364.
195-207 (1992)
Visual Field Differences in an Object Decision Task MELANIE Polytechnic
of East London,
VITKOVITCH
The Green, London,
El5 4L2 United Kingdom
AND GEOFFREY UNDERWOOD University of Nottingham,
Nottingham,
NG7 2RD United Kingdom
Two experiments are reported which investigate hemispheric processing in an object decision task. Experiment 1 used 40 pictures of objects, and 40 pictures of nonobjects, and subjects decided manually whether each lateralized stimulus represented an object. Results indicated an interaction between visual field and response (yes versus no). There was a right visual field advantage for positive responses, but no difference between visual field for negative responses. Positive responses were faster than negative responses, and this effect was more marked for right visual field presentations. These results were replicated in a second experiment. The results are interpreted as reflecting a left hemisphere superiority at accessing stored structural descriptions of known objects. The possibility that left and right hemispheres use different methods of carrying out the task is also discussed. 8 1992 Academic Press. Inc.
INTRODUCTION In the divided visual field literature, there is a relative lack of research on picture recognition, as compared to the numerous studies on word and face recognition. The present study examines hemispheric differences with respect to one particular aspect of object recognition. Young and Ellis (1985) have pointed out the need to relate data from studies of cerebral asymmetry to detailed theoretical models, and currently, there are a number of models of object recognition available (Humphreys, Riddoch, & Quinlan, 1988; Seymour, 1979; Snodgrass, 1984; Warren & Morton, 1982). These models generally suggest that three representational Reprint requests should be addressed to Melanie Vitkovitch, Department of Psychology, Polytechnic of East London, The Green, Romford Road, Stratford, London, El5 4L2 United Kingdom. 195 0278-2626192$5.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.
196
VITKOVITCH
AND UNDERWOOD
stages are involved in object recognition. Visual analysis of an object leads to a description of the physical features, and this input must then be matched against a stored structural description, so that the object may be recognized as a known object. Semantic information may then be accessed, and, as a third and final stage, the name of the object may be retrieved. It is not always possible to relate the results of visual field studies specifically to these models. An early conceptualization of the specialization of the two hemispheres was simply in terms of stimuli type, with the left hemisphere (LH) specialized for verbal stimuli and the right hemisphere (RH) specialized to deal with nonverbal stimuli. A number of studies (e.g., Bryden & Rainey, 1963; Dee & Fontenot, 1973) were designed around this conceptualization and compared the processing of nonverbal stimuli (such as geometric or nonsense shapes) with verbal stimuli (e.g., pictures of objects) in tasks such as shape matching. It was generally concluded that the ease with which a verbal label could be attached to a shape determines the visual field asymmetry, with easily labeled shapes, such as objects, leading to a right visual field advantage (RVFA), and geometric or nonsense shapes occasionally demonstrating a left visual field advantage (LVFA). Davidoff (1982) and Young and Ratcliff (1983) also state that the extent to which a form is named may be important. However, from these studies it is not easy to specify at which stage in the processing of objects any RVFA may be emerging. A task such as matching need not involve any stored representation stages. A further characterization of the two hemispheres was in terms of preferred modes of processing, rather than types of stimuli, and several dichotomies have been proposed, e.g., LH as an analytical processor and RH as a holistic processor. Some studies (e.g., Magnani, Mazzucchi, & Parmon, 1984; Tomlinson-Keasey & Kelley, 1979) have used the same pictures or shapes with different tasks that encourage one mode of processing or the other, and there is some support for an account which suggests that it is the processing demands of a task, rather than the nature of the stimuli, which may be a major determinant of cerebral asymmetry. In a similar vein, it has been suggested (e.g., Klatzky, 1972) that when picture stimuli are matched on a nominal basis, as opposed to a physical basis, then LH advantages should be apparent. The implication of this claim is that it is the process of attaching a label to the picture which leads to LH advantages, not the easily labeled picture per se. This latter idea might suggest that it may only be the final stage of object recognition (name retrieval) which leads to the LH advantage. In general, it is the case that the naming of lateralized picture stimuli leads to an LH advantage, possibly because of LH involvement in output processes (see Underwood & Whitfield, 1985), although this is not always the case (Young, Bion, & Ellis, 1980). It is generally accepted that the
LATERALIZED
OBJECT DECISION
197
RH is superior at the early stages of visuospatial processing (see Davidoff, 1982, and Young & Ratcliff, 1983, for reviews). The stages of recognition which, according to the models above, are thought to occur between visual processing of the stimuli, and their naming, have received much less attention. For example, there have been no studies of the normal population that, to our knowledge, have directly addressed the issue of hemispheric involvement in the earliest stage of object recognition (i.e., matching the visual input to a stored structural description). This contrasts with a recently accumulating body of research concerning RH involvement in the early stagesof face processing (e.g., Hay, 1981; Newcombe, De Haan, Ross, & Young, 1989; Rhodes, 1985; Young, Hay, & McWeeny, 1985; Young, Hay, McWeeny, Ellis, & Barry, 1985). In the clinical literature, there is one study which would suggest that the early stage of object recognition is lateralized to the RH (Warrington & Taylor, 1978). Young, Bion, and Ellis (1980), and Young and Bion (1981) do suggest, from their studies of picture naming in the normal population, that both hemispheres may be involved in the identification of objects, although they stress the need for further research to clarify the level at which RH identification of pictures is achieved. In the following two experiments, we use an object decision task in order to address the issue. The object decision task was first introduced by Kroll and Potter (1984) and bears some similarity to the lexical decision task (though see Kolers & Brison, 1984) and to face decision tasks. In the object decision task, nonobjects, constructed from combining parts of different true objects, are presented randomly with true objects. Subjects are required to decide whether the stimulus represents a true object or not. Logically, deciding whether a line drawing depicts a known object or not involves access to stored structural descriptions. Riddoch and Humphreys (1987), who use the task in a clinical context, maintain this point. Lupker (1988) also suggeststhat the task should tap the entry point to semantics, which from the model of object recognition outlined above is shown to be at the structural description stage. Lupker failed to find any semantic priming effects using the object decision task, suggesting that the task does not automatically involve semantic processing. Kroll and Potter argue that the objects are not being named during the task, and clearly no overt naming is required (avoiding the problem of LH speech output processes). There would seem to be some agreement that the task taps the early stage of processing, and for this reason, we investigate whether this task leads to lateral differences in the normal population. It is difficult from the preceding review of studies to make any firm predictions, although the RH superiority at visuospatial processing, and the studies by Warrington and Taylor, Young and Bion, and Young et al. suggest that the RH should not be at any particular disadvantage.
198
VITKOVITCH
AND
UNDERWOOD
EXPERIMENT 1 Subjects
Sixteen subjects took part in this experiment. All were undergraduates or postgraduates at the University of Nottingham, were right-handed by self-report, and used their right hand for writing in the normal position. All had normal or corrected-to-normal eyesight. Stimuli
Pictures of objects were selected from the standardized set of Snodgrass and Vanderwart (1980). For reasons related to other research in the same laboratory, 20 pictures of fruit and vegetables, and 20 pictures of manmade objects were used. In addition to the 40 positive stimuli, 40 pictures of nonobjects were selected from the set constructed by Kroll and Potter (1984). These nonobjects were formed by tracing parts of pictures of true objects and then regularizing the stimuli. The set of 40 nonobjects were carefully selected so that the true objects and nonobjects did not differ in any obvious way with respect to superficial characteristics. The stimuli were photographed (the nonobjects and objects were photographed to the same size) and mounted on slides. Two slides were prepared of each stimulus-one in each visual field. These slides were mirror images of each other. Apparatus and Display
Slides of pictures were displayed on a blank white screen by use of a Kodak Carousel back-projector. The shutter rise time was 1.5 msec. The projector was controlled by a Rockwell Aim 65 microcomputer, mediated by an interface. The illuminated field of the screen measured 9” by 5”. The illuminated level of the preexposure field was 45 candela/m2. The illuminated level of the exposure field was 286 candela/m2. A permanent fixation dot was centered on the illuminated screen. A chin rest, fixated to a table, was positioned in front of the back-projection screen such that, when the subject was seated, the distance from the back-projection screen was 42”. The visual angle from fixation point to the right side of the picture in the LVF and the left side of the picture in the RVF was 2”2’43”. The pictures were measured for vertical and horizontal angles subtended. The mean vertical angle was 2”2’43”, and horizontal angle was 2”34’40”. A small box containing the two response keys was positioned centrally in front of the chin rest and connected to the computer. The display procedure was as follows. A trial started with a warning tone. During this time the preexposure field was displayed. One second after the tone, the slide was displayed for an interval of 150 msec. With the termination of this display, the preexposure field reappeared. The subject made his/her response. There was then an interval of 1 set before
LATERALIZED
OBJECT DECISION
199
the next trial started, which would not commence until a response was made. Design and Procedure
The task required subjects to respond positively or negatively according to whether the displayed stimulus represented a real object or not. Stimuli were presented in either the LVF or RVF. There were therefore two within-subject factors-response type and field of presentation, each with two levels. The stimuli were presented in three blocks, with a short break in between (there were 54 experimental stimuli in two blocks, and 52 in a third block). The first eight trials in each block were practice stimuli. There were an equal number of positive and negative stimuli in each block, and an equal number of LVF and RVF stimuli. Apart from this, the stimuli were randomly allocated to blocks. Slides within each block were rerandomized every four subjects. The order of blocks was random. Subjects were instructed to fixate on a central dot at all times. They were told to press a “yes” key if the stimulus presented depicted a true object and to press the “no” key if the stimulus did not represent a true object. It was decided randomly whether subjects should use their index finger of the right hand for the positive key, and their left hand for the negative key, or vice versa. In addition to the practice stimuli within each block, subjects were given a practice session at the beginning of the session. They were instructed to respond as quickly and as accurately as possible. Results
The error rate in this experiment was found to be generally quite high, with a few subjects demonstrating very high error rates. A criterion of 85% correct responses (overall) was set, and any subjects with over 15% errors were excluded from the analysis. Five subjects were excluded from the analysis. Mean reaction times (RTs) were calculated for each subject for each condition, and the mean of these means for each condition is shown in Table 1. An analysis of variance indicated a main effect of response (F(1, 10) = 32.2, p < .OOl; MS, = 3065.17), with positive responses faster than negative responses, and an interaction between visual field and response (F(1, 10) = 9.48, p < .05; MS, = 699.42). There was no main effect of visual field (F < 1; MS, = 1290.46) The interaction was further analyzed using simple main effects (Kirk, 1968). For the positive responses, there was a reliable difference between LVF and RVF (F(1, 20) = 4.74, p < .05), with faster responses to RVF stimuli. There was no difference between LVF and RVF stimuli for the negative responses (F(1, 20) = 2.17,
200
MEAN
VITKOVITCH
AND
UNDERWOOD
TABLE 1 RTs (msec) AND PERCENTAGE ERRORS FOR THE POSITIVE AND NEGATIVE THE LATERALIZED
OBJECT DECISION
TASK IN EXPERIMENT
Positive
X RTs Error
RESPONSES IN
1
Negative
LVF
RVF
LVF
RVF
633 15%
604 13%
704 6%
723 11%
p > .05). The difference between positive and negative responses was significant for both the RVF (F(1, 20) = 41.57, p < 0.001) and LVF (F(1, 20) = 14.39, p =C.Ol), although inspection of the means shows a
much stronger effect in the RVF than the LVF. There was a positive skew to the RT data, and, accordingly, the data were submitted to a logarithmic transformation (Kirk, 1968; Howell, 1987). Analysis of the transformed data gave the same results, with respect to main effects and interaction and with respect to analysis of simple main effects. Errors were also analyzed. The main effect of response (more incorrect positive responses) and the interaction between response and VF just failed to reach significance (F(1, 10) = 4.90, p < .lO; MS, = 58.24) and (F(1, 10) = 4.72, < .lO; MS, = 30.81), respectively). The main effect of visual field was not reliable (F(1, 10) = 3.13, p > .lO; MS, = 9.41). The marginally reliable interaction reflects a reduced error rate for the LVF negative stimuli. Discussion
The analysis of positive and negative RT responses indicated a significant interaction between VF and response. Analysis of simple main effects revealed an RVFA for positive responses, but no significant VF difference for negative responses, although mean RTs were faster for LVF negative responses than RVF negative responses. Positive responses were overall quicker than negative responses, although this effect was more marked for RVF stimuli. The interaction between VF and response argues against any explanation for VF arising from an attentional bias in favor of one VF or the other, as suggested by theories such as Kinsbourne’s (1970). It is also the case that explanations in terms of lack of fixation, or scanning tendencies, are less relevant. These kinds of explanations would favor an overall VF advantage, unqualified by interactions with stimuli type (see Patterson & Besner, 1984; Young & Ellis, 1985; Zaidel & Schweiger, 1984). However, an interaction between VF and response type could possibly reflect a response bias for one hemisphere. For example, the quicker responses
LATERALIZED
OBJECT
DECISION
201
for positive stimuli in the RVF could reflect a tendency for the LH to respond “yes” readily. Similarly, the trend for faster negative response in the LVF might reflect an RH readiness to respond “no.” If this were the case, then error rates could be expected to reflect these kinds of bias. The analysis of errors indicated an interaction (marginally reliable) between VF and response. For the RVF responses, there was no difference in error rate for positive and negative responses, which is inconsistent with the suggestion that the faster RVF positive RTs were due to a response bias in favor of “yes” responses. If this were the case, error rates should be raised for “no” responses. For the LVF responses, error rates between positive and negative responses differed somewhat, with a reduced error rate for “no” responses. Furthermore, LVF negative responses were more accurate than RVF negative responses. This could possibly reflect a bias to respond “no” to LVF stimuli. By this account, one might expect more errors for positive LVF stimuli than positive RVF stimuli, and there is a slight increase in LVF error rates for positive stimuli, although it is not particularly marked. Given the possibility that a response bias cannot be completely ruled out from this error data, a replication study is reported which examines the reliability of the results. EXPERIMENT
2
This experiment is similar to Experiment 1, although only a subset of the stimuli were used, thereby shortening the session. The 20 fruit and vegetable stimuli only were used as objects, and accordingly, 20 nonobjects were chosen from the larger set. The nonobjects were again selected carefully so that any discrimination between this subset of objects and nonobjects could not be readily achieved by figural differences between the two stimuli sets. For example, the nonobjects had rounded or elongated shapes and their outlines were generally nonangular, to match the shapes of fruit and vegetables (e.g., apple, carrot). There was only one block of 80 stimuli, and these were preceded by a practice session. The apparatus and the procedure were the same as for Experiment 1. Twelve new subjects were tested, from the same population as the first experiment. Results The error rate in this experiment was less than in Experiment 1, although two subjects did not reach the criterion of 85% correct. These subjects were excluded from the analysis. The mean RT for each subject for each condition was calculated, and Table 2 presents the mean of these means. The data were analyzed as in Experiment 1. The main effect of response (faster positive responses) was marginally reliable (F(1, 9) = 3.94, p > .05; MS, = 18758.83). The main effect of VF was not reliable (F > 1;
202
VITKOVITCH
AND TABLE
MEAN
UNDERWOOD 2
RTs (msec)
AND PERCENTAGE ERRORS FOR THE POSITIVE AND NEGATIVE THE LATERALIZED OBJECT DECISION TASK IN EXPERIMENT 2 Positive
fi RTs Error
RESWNSES IN
Negative
LVF
RVF
LVF
RVF
685 7%
660 5%
754 9%
763 10%
MS, = 1221.12), but the interaction between VF and response was significant (F(1, 9) = 8.81, p < .05; MS, = 331.82). The interaction was analyzed using simple main effects. Positive responses were significantly faster than negative responses for RVF stimuli, but the effect of response for LVF stimuli was unreliable (F(1, 18) = 2.49, p > .05). The difference between RVF and LVF for the positive responses (faster positive responses) just failed to reach significance at the 5% level (F(1, 18) = 4.12, p > .05). For the negative responses, there was no difference between VF (F < 1). As in Experiment 1, the distribution of the data demonstrated a positive skew, and logarithmic transformation of the data and subsequent analysis gave the same pattern of effects as the raw data, with effects now reaching the 5% significance level. The main effect of response was significant (F(1, 9) = 5.88, p < .05), as was the interaction between VF and response (F(l) 9) = 9.45, p < .Ol). The main effect of VF was not significant (F < 1). The simple main effect of VF for positive responses was significant (F(1,18) = 5.49, p < .05), but the effect of VF for negative responses was not reliable (F < 1). The simple effect of response type for RVF was significant (F(1, 18) = 8.74, p < .Ol), and the effect of response type for LVF was marginally reliable (F(1, 18) = 3.28). As in Experiment 1, the effect of response was more pronounced for RVF stimuli as compared to LVF stimuli. Errors were also analyzed, but there were no effects (main effect of response (F(1, 9) = 2.31, p > .05; MS, = 2.12): main effect of VF (F < 1; MS, = .43) and the Response x VF interaction (F(1, 9) = 1.53, p > .05; MS, = 1.04)). Discussion The interaction between VF and response was replicated in this second experiment. The pattern of means in both experiments is the same, with RVF positive responses faster than LVF positive responses. There was no difference between LVF and RVF negative responses, although again, as in Experiment 1, LVF responses were slightly faster. The difference between positive and negative responses was significant for RVF responses. LVF positive responses do show faster response times than LVF
LATERALIZED
OBJECT DECISION
203
negative responses, though in both experiments, this effect is not as marked as for RVF responses. The error analysis in this second experiment, however, did not show any significant differences between conditions. This suggeststhat response bias does not account for the interaction between VF and response type in the RT analysis. It can be noted that error rates in this second study are less than in the first experiment. This could be due to the shorter experimental session, although, given that the reduction in errors is shown mainly for the positive stimuli, it is more likely that the number of errors in Experiment 1 were due to particular stimuli not present in Experiment 1. In fact, inspection of the errors does indicate that subjects tended to consistently make errors to certain stimuli, suggesting these stimuli were not easily recognizable in the context of the present task. GENERAL DISCUSSION In two experiments, we examined visual field differences in an object decision task. Subjects were required to judge whether lateralized stimuli represented true objects or not. There was some difficulty in performing this task, evident by relatively high error rates. At least part of this difficulty was due to certain objects which, in the context of the nonobject fillers, were apparently not very easy to recognize. Nevertheless, an interesting pattern of results emerged, and this was consistent over both experiments. Analysis of the RTs to lateralized true objects and nonobjects indicated both a main effect of response and an interaction between visual field and response type (“yes” versus “no”). The pattern of means demonstrates that positive responses to true objects were faster than negative responses to nonobjects. This result is consistent with data from object decision tasks performed under central viewing (Kroll & Potter, 1984; Lupker, 1988). It suggests generally that the task is not being performed on any superficial physical characteristics which would allow an easy discrimination between true objects and nonobjects, without accessing stored structural descriptions. Were this the case, nonobjects could be rejected as quickly as true objects are accepted. However, the RT data for the LVF presentation of stimuli did show a less marked difference between true objects and nonobjects than the RVF data. The interaction between visual field and response type may be an indication that the LH and RH carry out the task in different ways, In the Introduction, it was noted that there have been suggestions that the LH analyzes stimuli in an analytical manner, the RH in an holistic fashion. The more pronounced difference between positive and negative responses for RVF stimuli may reflect, for example, an analytical feature comparison process adopted by the LH. If the LH analyzes the stimulus in terms of features, then partial activation may occur in stored structural descriptions
204
VITKOVITCH
AND UNDERWOOD
which contain those features. This would occur for both true objects, and nonobjects, given that the latter are constructed from the parts of true objects (see under Stimuli). Top-down processing could speed the recognition of true objects, but a time-consuming checking process may have to be implemented in order to reject the nonobjects. In contrast, the RH may analyze the stimuli in a more holistic fashion. Negative responses, since they do not as a whole match any true object, may be readily rejected, and hence positive and negative RTs do not differ so markedly for LVF stimuli. These suggestions could be tested by varying the degree of featural similarity between nonobjects and objects. Increasing the similarity between objects and nonobjects would then be expected to influence the responses to RVF stimuli, without affecting LVF responses. In both experiments, an RVFA was found for positive responses. There was also a slight tendency for faster responses to LVF nonobject stimuli (consistent with the studies mentioned in the Introduction which suggest LVFA for nonsense stimuli). However, the VF differences for negative responses were not significant. An absence of VF effects for negative responses is not uncommon in VF studies (Young, Hay, & McWeeny, 1985), and some dismiss this as inconsequential, providing that the positive responses demonstrate consistent advantages. However, Lambert (1982) provides a fuller discussion of the possible interpretations of VF advantages, which we consider here in the context of our own results. We have already argued that explanations such as attentional bias and scanning tendencies are unlikely in the present case (see Discussion to Experiment 1). With respect to hemispheric differences, an RVFA may be an indication that the processing of stimuli and decisions relevant to the task may be carried out by both hemispheres, with the LH simply more efficient at so doing (whether this is a result of similar methods used by both hemispheres, or different methods, as suggested above). In terms of the theoretical model of object recognition outlined, our data from the object decision task would suggest that the LH is better at either accessing stored structural descriptions, and/or at deciding a given stimulus corresponds to a relevant stored description. (Given the accepted RH superiority at visuospatial processing generally, it would seem unlikely that the LH superiority arises from an earlier visual analysis stage, or that it is due to the construction of an object description.) Alternatively, the RVFA may indicate that only the LH can carry out the processing relevant to the object decision task. The RVFA may then reflect the fact that RVF stimuli have more direct accessto the LH areas of the brain which carry out the processing. That is, the RVFA may be an indication that information from LVF stimuli may have to be transmitted to the LH (via the corpus callosum) for processing. In the present case, this would suggest the possibility that only the LH possessesstructural descriptions for known objects. However, the absence of VF asym-
LATERALIZED
OBJECT DECISION
205
metry for negative responses would then imply that the RH was capable of carrying out negative responses to nonobjects. As Lambert (1982) points out, this does not seem plausible. A further theoretical possibility considered by Lambert is that both LVF objects and LVF nonobjects may be transmitted via the corpus callosum to the LH. The interaction of VF with response would then imply that the LH processing of objects and nonobjects is differentially affected according to whether the input route is direct or indirect. Thus, the LH can process nonobjects equally rapidly, whether the stimulus is received directly via the RVF or indirectly from the LVF. True objects are processed more quickly via the direct route. Again, this explanation also seems unlikely. For the present, since we cannot determine the precise explanation for the RVFA for recognizing true objects, we conclude generally that the LH is superior at recognizing objects, whether it has a more efficient method than the RH or it is as a result of LH specialization. The results of the present study contrast with lateralized studies of face recognition which have used a face decision task. For example, in the study by Young, Hay, and McWeeny (1985), subjects had to decide if a presented stimulus depicted a true face or not. Stimuli were either line drawings of faces, or they were either moderately or highly scrambled faces. Young et al. found an LVFA when the scrambled faces resembled more closely true faces, and they concluded that the RH may be more involved in the construction of a detailed representation of a face. This would seem to support the conclusion drawn above that it is more likely that the LH advantage found in the object decision task study is due to processes which occur after the construction of the object description. A second possibility is that the LVFA found for the face decision task may be due to processing which may be unique, or more specific to faces. It has been suggested that the processing of faces may require configurational processing of the parts of the faces (Rhodes, 1985) and that this may favor RH processing. It can be noted that the contrasting set of results to the similar lateralized face and object decision tasks add strength to our argument that the present results are unlikely to be due to response bias, but genuinely reflect differences in hemispheric processing. It is of particular interest that the hemispheric processing of objects and faces would appear to differ at the early stages of recognition. This may be due to the nature of the processing required for the two different types of stimuli. A single experiment comparing face and object decisions, varying the nature of the nonobjects and nonfaces would be informative. In conclusion, the present data suggest an LH superiority at recognizing true objects in an object decision task. We suggest that this advantage is due to processesinvolving accessto, or comparison with, stored structural descriptions of objects. LH and RH processing of positive and negative
VITKOVITCH
206
AND UNDERWOOD
responses are shown to differ, and this could be an indication of different methods of performing the task. Further research on object decisions could follow the work on lateralized face decisions and examine the effect on visual field differences of varying systematically the degree of overall likeness between nonobjects and true objects. APPENDIX Positive stimuli used in Experiments 1 and 2: Ashtray, axe, barrel, balloon, bell, bowl, button, comb, cup, doorknob, drum, glass, guitar, pen, rolling-pin, shoe, thimble, top, wheel, whistle, apple, artichoke, asparagus, banana, carrot, celery, cherry, corn, lemon, mushroom, onion, orange, peach, pear, pepper, pineapple, potato, pumpkin, strawberry, tomato. (Only the last 20 fruit and vegetables were used in Experiment 2).
REFERENCES Bryden, M. P., & Rainey, C. A. 1963. Left-right differences in tachistoscopic recognition. Journal of Experimental
Psychology,
66, 568-571.
Davidoff, J. 1982. Studies with non-verbal stimuli. In J. G. Beaumont (Ed.), Divided visual field studies of cerebral organisation. London: Academic Press. Dee, H. L., & Fontenot, D. J. 1973. Cerebral dominance and lateral differences in perception and memory. Neuropsychologia, 11, 167-174. Hay, D. C. 1981. Asymmetries in face processing: Evidence for a right hemisphere perceptual advantage. Quarterly Journal of Experimental Psychology, 33A, 267-274. Howell, D. C. 1987. Statistical methodr for psychology. Boston: PWS-Kent. Humphreys, G. W., Riddoch, M. J., & Quinlan, P. 1988. Cascade processes in picture identification. Cognitive Neuropsychology, 5, 56-103. Kinsboume, M. 3970. The cerebral basis of lateral asymmetries in attention, Acta Psychologica, 33, 193-201. Kirk, R. E. 1968. Experimental design: Procedures for the behavioural sciences. California: Brooks/Cole. Klatzky, R. L. 1972. Visual and verbal coding of laterally presented pictures. Journal of Experimental
Psychology,
%, 439-448.
Kolers, P. A., & Brison, S. J. 1984. Commentary: On pictures, words and their mental representations. Journal of Verbal Learning and Verbal Behaviour, 23, 105-113. Kroll, J. F., & Potter, M. C. 1984. Recognizing words, pictures and concepts: A comparison of lexical, object and reality decisions. Journal of Verbal Learning and Verbal behaviour, 23, 39-66. Lambert, A. J. 1982. Right hemisphere language ability. 2. Evidence from normal subjects. Current Psychological
Reviews, 2, 139-152.
Lupker, S. J. 1988. Picture naming: An investigation of the nature of categorical priming. Journal of Experimental
Psychology:
Learning,
Memory
and Cognition,
14, 444-455.
Magnani, G., Mazzucchi, A., & Parmon, M. 1984. Interhemispheric differences in same versus different judgments upon presentation of complex visual stimuli. Neuropsychologia,
22, 527-530.
Newcombe, F., De Haan, E. H. F., Ross, J., & Young, A. W. 1989. Face processing, laterality and contrast sensitivity. Neuropsychologia, 27, 523-538.
LATERALIZED
OBJECT DECISION
207
Patterson, K., & Besner, D. 1984. Is the right hemisphere literate? Cognitive Neuropsychology, 1, 315-341. Rhodes, G. 1985. Lateraked processes in face recognition. British Journal of Psychology, 76, 249-271. Riddoch, M. J., & Humphreys, G. W. 1987. Visual object processing in optic aphasia: A case of semantic access agnosia. Cognitive Neuropsychologia, 4, 131-185. Seymour, P. H. K. 1979. Human visual cognition. London: Collier Macmillan. Snodgrass, J. G. 1984. Concepts and their surface representations. .lourna[ of Verbal Learning and Verbal Behaviour,
23, 3-22.
Snodgrass, J. G., & Vanderwart, M. A. 1980. Standardized set of 260 pictures: norms for name agreement,image agreement, familiarity and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174-215. Tomlinson-Keasey, C., & Kelley, R. R. 1979. A task analysis of hemispheric functions. Neuropsychologia, 17, 345-351. Underwood, G., & Whitfield, A. 1985. Right hemisphere interactions in picture-word processing. Brain and Cognition, 4, 273-286. Warren, C. & Morton, .I. 1982.The effects of priming on picture recognition. British Journal of Psychology, 73, 117-129. Warrington, E. K., & Taylor, A. M. 1978. Two categorical stages of object recognition. Perception,
7, 696-705.
Young, A. W., & Bion, P. J. 1981. Identification and storage of line drawings presented to the left and right cerebral hemispheres of adults and children. Cortex, 17, 459-463. Young, A. W., Bion, P. J., & Ellis, A. W. 1980. Studies toward a model of laterality effects for picture and word naming. Brain and Language, 11, 54-65. Young, A. W. & Ellis, A. W. 1985. Different methods of lexical accessfor words presented in the left and right visual hemifields. Brain and Language, 24, 326-358. Young, A. W., Hay, D. C., & McWeeny, K. H. 1985. Right cerebral hemisphere superiority for constructing facial representations. Neuropsychologia, 23, 195-202. Young, A. W., Hay, D. C., McWeeny, K. H., Ellis, A. W., & Barry, C. 1985. Familiarity decisions for faces presented to the left and right cerebral hemispheres. Brain and Cognition, 4, 439-450. Young, A. W., & Ratcliff, G. 1983. Visuospatial abilities of the right hemisphere. In A. W. Young (Ed.), Functions of the right cerebral hemisphere, London: Academic Press.
Zaidel, E., & Schweiger, A. 1984. On wrong hypotheses about the right hemisphere: Commentary on K. Patterson & D. Besner “Is the right hemisphere literate?” Cognitive Neuropsychology,
1, 351-364.