Famous personal names and the right hemisphere: the link keeps missing

Famous personal names and the right hemisphere: the link keeps missing

Brain and Language 82 (2002) 95–110 www.academicpress.com Reply Famous personal names and the right hemisphere: the link keeps missingq Stefan R. Sc...

150KB Sizes 0 Downloads 13 Views

Brain and Language 82 (2002) 95–110 www.academicpress.com

Reply

Famous personal names and the right hemisphere: the link keeps missingq Stefan R. Schweinberger,* J€ urgen M. Kaufmann, and Andy McColl Department of Psychology, University of Glasgow, 58 Hillhead Street, Glasgow G12 8QB, UK Accepted 12 February 2002

Abstract In this reply to the comment by VanLancker and Ohnesorge (2002), we present the case that current evidence supports the role of left hemisphere in the recognition of famous personal names. We argue that this conclusion is in line not only with the results of Schweinberger, Landgrebe, Mohr, and Kaufmann (2002), but also with the evidence from methods other than divided visual field studies (e.g., PET and ERP studies). We show that our view is also supported by a new set of experiments that address a major concern raised by VanLancker and Ohnesorge in their comment and discuss why the evidence they present does not provide conclusive support to their right hemisphere hypothesis. One of the several possible reasons for this failure is that famous name stimuli may be less suitable than personally familiar stimuli to elicit personally relevant, affective aspects of recognition. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Personal names; Hemispheric asymmetries; RT

1. Introduction We welcome the opportunity to reply to the comment by Diana VanLancker and Clark Ohnesorge (VanLancker et al., 2002)—subsequently the label ‘‘V&O’’ will be used as a convenient shorthand to our recent paper (Schweinberger et al., 2002; subsequently: ‘‘SLMK’’). We will also comment on their recent paper (Ohnesorge & VanLancker, 2001, subsequently: O&V) which is relevant for this discussion. After a detailed reading of both the paper by O&V and the comment on our paper by V&O, we wish to maintain that the link between personal names and the human right hemisphere (RH) keeps on missing. Before discussing the evidence in detail, we find it useful to distinguish at least two versions of their RH hypothesis. According

q We gratefully acknowledge the help of Esther Pickering in obtaining the familiarity ratings for the names used in this study. * Corresponding author. Fax: +44-0-141-330-4606. E-mail address: [email protected] (S.R. Schweinberger).

0093-934X/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 0 0 9 3 - 9 3 4 X ( 0 2 ) 0 0 0 3 5 - 4

96

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

to a strong version of the RH hypothesis, the RH would actually be superior to the left hemisphere (LH) in the processing of personal names. This strong version of the RH hypothesis was advanced in earlier publications by Diana VanLancker and her colleagues (e.g., VanLancker & Klein, 1990; VanLancker, Klein, Hanson, Lanto, & Metter, 1991). According to a weak version of the RH hypothesis, while LH performance would be as good as or even superior to that of the RH, the LH superiority for names would be smaller compared to other word classes such as common nouns thereby suggesting a relatively greater contribution of the RH to processing in the case of names. It is this weaker version of the RH hypothesis that was advanced in the current papers by O&V and V&O. In our view, convincing evidence has been presented neither for the strong nor for the weak version of the RH hypothesis. Our arguments for that conclusion can be broadly classified into four categories which we consider in turn. These are (a) the scope of our original criticism and the consideration of data from methods other than divided visual field studies, (b) an evaluation of the evidence presented recently by O&V, (c) an evaluation of the arguments put forward by V&O in response to the paper by SLMK, and (d) the presentation of data from a set of new experiments.

2. Hemispheric asymmetries for personal names investigated with methods other than divided visual field presentation This first point relates to the scope of our original criticism. V&O understand our paper as a refutation of their conference paper (Ohnesorge et al., 1999). Their suggestion that we should have attempted a more precise replication of their findings would seem reasonable, had our criticism been targeted at that paper specifically. However, we submit that we made it quite clear that our criticism was not focused on that conference paper alone, but more generally challenged the hypothesis that the RH has a special role in personal name recognition. Experiments with the divided visual field methodology can contribute to an evaluation of this RH hypothesis, but clearly a consideration of the relevant findings obtained with other methods (e.g., PET or ERP data, investigations in brain-lesioned patients) is required as well. Evidence from functional brain imaging strongly implicates the role of the LH but not the RH in famous name recognition; this is evidence which V&O either discount or fail to fully appreciate. For example, O&V discount a recent PET study that reports LH activation but fails to find RH involvement in name processing (Gorno-Tempini et al., 1998), on the basis of the use of a compound subtraction methodology which they consider problematic. We could not agree more with a careful and critical approach to subtraction methodology in neuroimaging. However, O&V appear to be more lenient towards other data from neuroimaging that, at least on the face of it, appear to support the RH hypothesis. For instance, they claim that a role of the RH in proper noun recognition ‘‘is consistent from lesion and functional scanning evidence’’ (p. 150) and in support of this claim they specifically quote a PET study by Damasio, Grabowski, Tranel, Hichwa, and Damasio (1996). Unfortunately, a closer look at that article reveals that this quotation is misleading. In the Damasio et al. study, brain activation was studied in naming tasks involving faces, animals, and tools. Personal names were produced in response to faces, whereas basic object level names were produced in response to other categories. V&O regard the finding that naming faces produced significant activation of both left and right temporal poles, as support of their RH hypothesis. However, the activation for familiar face naming was observed relative to a control

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

97

task that involved unfamiliar faces and thus did not subtract activation generated by the inevitable recognition of each face before naming. Given that the role of the RH in familiar face recognition is extremely well established, the most plausible explanation for the RH activation in the Damasio et al. paper is that it is related to familiar face recognition, rather than to name production. Critically, this design problem was explicitly identified in the original paper by Damasio et al.—a point which O&V missed. To quote from the original paper, ‘‘. . .the person-naming task also activated right TP. The control task used unfamiliar rather than familiar faces, and thus did not subtract activity generated by the inevitable recognition of each person before naming. These findings emphasize the difficulty of separating recognition and naming processes. . .’’ (Damasio et al., 1996, p. 503). We conclude that the quotation by O&V of neuroimaging evidence that favors the RH hypothesis is misleading and does not withstand a careful reading of that study. In contrast, there is consistent evidence from neuroimaging for LH involvement in name processing. To the study of Gorno-Tempini et al. (1998) we add further neuroimaging studies that also contradict the RH hypothesis. A PET study by Sergent and co-workers (Sergent, MacDonald, & Zuck, 1994) found strong evidence for LH involvement in name recognition and it did so without using a compound subtraction methodology that O&V criticized. A very recent PET study (Grabowski et al., 2001) also found a strong involvement of left anterior temporal regions in name retrieval, with the additional intriguing finding that retrieving personal names and names of unique landmarks appeared to activate the same regions in the left temporal pole. Moreover, V&O also fail to consider work from event-related brain potentials (e.g., Dehaene, 1995; Proverbio, Lilli, Semenza, & Zani, 2001) which is not subject to critique based on subtraction methodology. In line with the PET evidence, these ERP studies also confirm the strong LH involvement in name processing and some even suggested a larger LH asymmetry for proper names as compared to common nouns (Proverbio et al., 2001).1 Findings in brain-lesioned patients might also lend support on the right hemisphere hypothesis. In fact, it appears that the RH hypothesis was originally triggered by rather striking findings in brain-lesioned patients. First, globally aphasic patients may show preserved recognition of personal names but not other words in pictureto-name matching tasks (VanLancker & Klein, 1990). This is an intriguing finding which has also been replicated by others. For instance, Warrington and Clegg (1993) found preservation of proper place names (but not personal names) in an aphasic patient. These authors also suspected a contribution of the RH, although they were careful enough to mark this explicitly as a speculation. Second, relative to LH lesions, RH lesions caused larger deficits in a task that required pointing to a face that corresponded to a spoken name (VanLancker et al., 1991). Since we have previously argued in detail (Schweinberger, 1995; Schweinberger et al., 2002) why these findings are not conclusive evidence for the RH hypothesis, we will not repeat these arguments here.

1

It should be added that this study was concerned with retrieval and production (rather than recognition) of names. Also, to be fair V&O can hardly be blamed for not considering this very recent paper, which may not yet have been available by the time they wrote their reply. It hardly needs to be mentioned that all these PET and ERP studies provided more specific information about those areas within the LH that were involved in the processing of personal names. However, a detailed discussion of this evidence would be beyond the scope of this paper, which is only concerned with hemispheric asymmetries (but see, e.g., Semenza, Mondini, & Zettin, 1995).

98

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

It is useful to remember that the LH is dominant for most language functions, but the RH is not just a silent partner in language comprehension. RH contributions to language and word recognition have been shown in both healthy participants and split-brain patients, although the RH seems to be more proficient in comprehension than production (Gazzaniga, 1983; Nass & Gazzaniga, 1987; Schweinberger, Sommer, & Stiller, 1994; Zaidel, 1983). Even a weak version of the RH hypothesis would therefore require evidence that the degree of RH contribution to personal name recognition exceeds that for the recognition of other word classes. In the following, we will argue in detail why we think of the data presented by O&V and V&O rather than providing such evidence, cast further doubt on the RH hypothesis.

3. The data presented by O&V and V&O First, we note that no single one out of the eight experiments (six experiments in O&V plus two experiments in V&O) yielded an advantage of the LVF/RH for personal names. At the most, a modulation of an advantage for the RVF/LH is found in some conditions. This appears to rule out a strong version of the RH hypothesis (VanLancker & Klein, 1990; VanLancker et al., 1991). Second, although a reduced RVF/LH advantage for names relative to common nouns might support a weak version of the RH hypothesis, such a pattern of results is also not consistently found across their experiments. A short summary of O&V’s results, as we understand them, is as follows. In Experiment 1, proper names yielded a smaller, and nonsignificant, RVF/LH advantage, compared with common nouns. In Experiment 2, proper names and common nouns were presented in separate blocks; male/female and animate/inanimate decisions were required, respectively. This time, proper names yielded a larger RVF/LH advantage, compared with common nouns. In Experiment 3, which was a version of Experiment 2 with slightly changed exposure times, a larger RVF/LH advantage for names was found again.2 In Experiment 4, familiar and unfamiliar personal names were shown and the task was to judge the familiarity of each name. An RVF/LH advantage of similar magnitude was found for familiar and unfamiliar names. In Experiments 5 and 6, a new set of stimuli was produced and the task was to judge the familiarity of famous and unfamiliar names. Experiment 5, with 80 ms presentation time, yielded no significant hemispheric differences. Experiment 6, with 93 ms presentation time, yielded a significant RVF/LH advantage and a just significant interaction with familiarity, reflecting the finding that the RVF/LH advantage was strong and significant for unfamiliar names, but weak and nonsignificant for familiar names. On reading the paper by O&V, it was difficult for us to see what theoretical considerations guided their variations across experiments of the stimuli, presentation modes, and tasks. Overall, 2 studies (Experiments 1 and 6) provided support for a weak version of the RH hypothesis, 2 studies (Experiments 4 and 5) produced inconclusive results, and 2 studies (Experiments 2 and 3) produced results that

2

In Experiments 2 and 3, although this significant interaction between hemisphere and word type can be seen in their Fig. 1, we note that O&V do not explicitly state in the text that the LH advantage was significantly larger for names than common nouns. Instead, they compare the accuracy between names and nouns within each hemisphere. To us this particular comparison appears tangential for the study of hemispheric differences and of minor relevance because word types were not matched for overall difficulty. In the description of results, we have therefore focused on hemispheric differences which we considered to be of primary interest both for their study and this debate.

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

99

contradict the RH hypothesis. We were surprised by how, through this rather inconsistent pattern of results, O&V identified clear support for the RH hypothesis. In the abstract, O&V emphasize the importance of stimulus familiarity and task factors, stating that performance for famous names in the two visual fields did not differ when the task required categorization into famous and unfamiliar. However, this statement is based mainly on Experiments 5 and 6 and does not explain why similar findings were obtained in Experiment 1 with a very different task. Unfortunately, even results of Experiments 5 and 6 are inconclusive, as a consequence of the nature of stimulus selection used by O&V. The set of stimuli used in Experiments 5 and 6 differed strongly from that in O&V’s previous experiments, that had used only personal names as the proper name stimuli. In Experiments 5 and 6, familiar names could be personal names, names of comic figures, geographical place names, country names, buildings, brand names, etc. However, some subtypes of proper names may have different processing characteristics. For instance, country names can act as adjectives (‘‘British,’’ ‘‘French’’), whereas personal names do not normally take adjectival form. These subtypes of proper names can also be dissociated on the basis of experimental manipulation, which is why it has been claimed that personal names may be pure referring expressions but country names are not (Hollis & Valentine, 2001). Related to this, inspection of the stimulus set used by O&V and V&O (see O&V, Appendix D) suggests that their familiar and unfamiliar names were not remotely matched for the frequency of personal names and other subtypes of names. In the discussion of methodological issues below, we will explain how this odd aspect of stimulus selection may have contributed critically to the results they obtained.

4. The discrepancy between O&V and Schweinberger et al. (2001): methodological issues V&O identified several factors that might account for different results of their experiments and ours. Specifically, they mentioned as potential factors (a) the fact that our famous names had been selected by the experimenters and had not been subjected to rigorous familiarity ratings, the most important factor in their view, (b) our use of capital letters, (c) our presentation with first name centered above surname, (d) the fact that our studies involved time pressure, (e) our presentation of a string of ‘‘X’’-characters in the contralateral hemifield, (f) our exposure times, and (g) the fact that responses were carried out with both hands. They were unsure exactly how factors (d)–(g) might have influenced our results. However, they advanced the specific concern that insufficient familiarity with our names might have obscured a role of the RH (factor (a)). To a smaller extent, similar concerns were raised for factors (b) and (c): V&O argue that both our presentation of names in uppercase font and our format (with first name shown above surname) deviate from normal reading and may have reduced the perceived familiarity of the stimuli. Although this is possible, we note that recognition rates for famous names in our study were typically as good as or even better than those reported by O&V and V&O. The differences in accuracy between unfamiliar and famous names that V&O noted may simply be a result of response strategies. It is likely that instructions in our RT task differed from theirs and it is well known that differences between two response alternatives in both speed and accuracy can easily be caused by relatively subtle differences in instruction-induced response strategies (Ratcliff, 1985). Apart from this comment, we will not discuss in detail V&O’s most serious concern about the degree of familiarity of the names we used, but rather report a new study that

100

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

considered V& O’s suggestion to first obtain familiarity ratings from a cohort of our experimental subjects and used only famous names that elicit high familiarity ratings. The possibility that presentation of our names in uppercase font could affect the perceived familiarity cannot be easily rejected. However, we note that in their own experiments, V&O did not seem to be too concerned about the uppercase font. They used the uppercase font in Ohnesorge and VanLancker (1999), as well as in four out of six experiments in O&V (Experiments 1, 4, 5, and 6).3 As uppercase font was used in those two out of six experiments that provided some support for the RH hypothesis (Experiments 1 and 6), it appears difficult to argue that our use of uppercase font interfered with RH processing. We also appreciate the possibility to discuss in detail the issue of presentation format. Our format (with first name shown above surname), although not uncommon in experimental presentation of names, clearly deviates from normal reading. Although we think it is unlikely that this would greatly affect the perceived familiarity, this possibility again cannot be easily rejected. We still believe that our presentation format is valid, in particular, when using the divided visual field technique. We hold in turn that the horizontal presentation format used by V&O creates far more serious problems. Specifically, the surname part (e.g., ‘‘Bush,’’ ‘‘Clinton’’) in most personal names carries more information with respect to identity than does the first name part (e.g., ‘‘George,’’ ‘‘Bill’’). Presenting names in horizontal format means that for LVF/RH trials, the more informative part of the stimulus falls closer to the fixation point, compared with RVF/LH trials. Because visual acuity falls off dramatically and approximately linearly with increasing distance from fixation (Anstis, 1974), one would predict that this causes a systematic bias in favor of the LVF/RH for horizontally presented personal names. The very finding which V&O would prefer to attribute to a special RH ability could therefore be a consequence of their presentation format and it is this potential confound which our presentation format avoids. A close inspection of the set of stimuli (see Appendix D in O&V) that were used both in Experiments 5 and 6 by O&V and Experiments 1 and 2 by V&O suggests that this concern is even more serious. These experiments (at least two of them) found a reduced RVF/LH advantage for famous names. However, they used famous names that included personal names but also names of comic figures, geographical place names, countries, buildings, brand names, etc. Unfortunately, famous names and unfamiliar ‘‘matches’’ differed vastly with respect to the frequency of these name categories. For example, 57/100 famous names were clearly personal names, whereas only 21/100 unfamiliar matches were personal names. Instead, most unfamiliar items were compound words (e.g., ‘‘ping pong,’’ ‘‘banjo player,’’ ‘‘hearing loss,’’ ‘‘cinnamon roll’’). Thus, famous and unfamiliar items differed in important respects other than familiarity or ‘‘notoriety.’’ More worrying, this imbalance between the familiar and unfamiliar sets could completely explain the reduced RVF/LH advantage for familiar as opposed to unfamiliar names. Here is our explanation. For a majority of unfamiliar stimuli (e.g., compound words), first and second words were equally informative on average and thus the conventional RVF/LH advantage for word

3

The reader may note occasional inconsistencies in the description of the font type which O&V and V&O used for their experiments. For instance, O&V indicate both in their Table 1 and in the text on p. 146, that uppercase font was used for all stimuli in Experiments 5 and 6. In contrast, V&O say that title case font had been used in these experiments. For the purpose of the present discussion, we have assumed that the description in the original paper was correct.

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

101

recognition was found. For a majority of familiar stimuli (personal names), the more informative second word was closer to fixation and could therefore be seen better in the case of LVF/RH trials thereby causing a bias in favor of the RH performance. This is a serious concern which requires clarification, and we suggest that O&V attempt a replication of their findings with sets of familiar and unfamiliar names which are carefully matched to avoid this problem.

5. New experiments We will now report the results of two new experiments that both deal with concerns that V&O raised with respect to our recent findings and that consider their explicit suggestions for further study (V&O, p. 128—editor, please note: page number refers to the drafted manuscript; please replace as deemed fit). Relative to our original experiments, these involve the following alterations: first and most important, we now obtained familiarity ratings for all famous personal names from a cohort of our experimental subjects and ensured that all stimuli were highly familiar by using only those famous names that elicited high familiarity ratings. Second, one remaining concern with our previous paper may have been that performance levels for personal names were somewhat lower, compared to common nouns. By choosing very highly familiar names we now made an attempt to equate the overall task difficulty. Third, V&O were concerned that the presentation of a string of X-characters in the visual field contralateral to the stimulus (which we had used to reduce the tendency for involuntary eye movements towards a peripheral stimulus) might have interfered with stimulus processing. In addition to the unilateral conditions in which a stimulus was accompanied by a contralateral string of X-characters, we now added a bilateral condition that omitted the string of X-characters and instead presented identical copies of the stimulus to both visual fields (BVF) simultaneously.4 Fourth, while our original results had been obtained with German, participants, a move from Konstanz, Germany to Glasgow, Scotland, enabled us to establish that our previous findings were independent of language (at least as far as English and German are concerned). This is a small but not necessarily trivial extension of earlier findings, as personal names and words may differ across languages, e.g., in terms of the relationship between orthography and phonology etc. Finally, gender differences in functional hemispheric asymmetries may exist (McGlone, 1980). This point has not been raised by V&O as a possible reason for the discrepant findings nor is it particularly clear how exactly gender differences would influence results in the context of present studies. We simply note that gender differences do not appear to have been analyzed in either study and we therefore evaluated whether gender affects hemispheric asymmetries for personal names. 5.1. Experiment 1 5.1.1. Method Participants. Twenty participants (10 women) aged 19–43 (M ¼ 26:3 years) contributed data to this study. All participants were strongly right-handed (handedness

4 We originally added this condition to investigate the possible hemispheric collaboration (see e.g., Mohr, Pulverm€ uller, & Zaidel, 1994). However, for the purpose of this paper, this manipulation may also provide some information with respect to possible interfering effects of the string of X-characters in the contralateral hemifield.

102

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

scores > þ60 in an adapted version of the Edinburgh handedness questionnaire (Oldfield, 1971), M ¼ þ95), were native English speakers, and reported normal or corrected-to-normal visual acuity. Stimuli and apparatus. The words were 32 nouns of low frequency (<50) per million according to Johansson, M ¼ 5:9 (Johansson, 1989). The pseudowords were 32 pronounceable letter strings, each of which was matched to one word in that the same letters were used but in a rearranged fashion. All words or pseudowords consisted of two syllables and between 4 and 7 letters, presented in uppercase (see Appendix A). A computer monitor was used to display these stimuli in white on a dark gray background with the IBM8BIT font at font size 18, corresponding to a letter height of 1.1 cm. The stimulus width was between 3.0 and 5.8 cm, depending on the word length, corresponding to a visual angle of between 1.7° and 3.3° at a viewing distance of approximately 1 m. The stimuli were presented at a lateral eccentricity of 5.0 cm (measured from the fixation point to the stimulus center), corresponding to a visual angle of 2.9°. Procedure. Participants were informed that they would be shown pronounceable letter strings and that they should perform one of the two different key press responses, depending on whether or not the letter string represented a legal English word. Speed and accuracy were emphasized. For words, they should press the F1 and F12 keys of a computer keyboard with both the middle fingers. The correct response for pseudowords was a simultaneous key press of the F2 and F11 keys with both index fingers. For practical reasons, the keyboard was turned by 180°. A fixation cross was presented at the center of the monitor and participants were encouraged to focus this at all times. In every trial, a word or pseudoword was shown either in the LVF, the RVF, or simultaneously in BVF for 150 ms, while the fixation cross was still visible. For unilateral trials only, a string of X-characters was simultaneously shown at the screen location contralateral to the target stimulus, with the average number of characters corresponding to the average number of letters in the target stimuli. This was done to increase the likelihood that attention would be focused centrally and also to reduce or eliminate the tendency to perform involuntary eye movements to unilateral stimulation. Response times were accepted within a window of 150–1800 ms after stimulus onset. In trials with incorrect response, a feedback tone (150 ms, 500 Hz) was presented after 2000 ms; for trials in which no response had been made within 1800 ms, a different feedback tone (150 ms, 650 Hz) was presented. The intertrial interval between target stimulus onsets was either 4000 ms or 4150 ms (depending on whether or not a feedback tone had been presented in between trials). A total of 192 target letter strings (32 words and 32 pseudowords, with each stimulus presented once in each of the three visual field conditions) was shown. A break was allowed after 64 experimental trials and 18 practice trials were conducted before the experiment. The stimuli in these practice trials were not used subsequently. Responses were scored as correct, if the appropriate pair of keys was pressed within the specified time window (150–1800 ms). For the calculation of reaction times (RTs), the first key that was pressed was always considered. Errors of omission (no key press), errors of commission (wrong pair of keys), and inconsistent responses (trials in which the computer registered two key presses that did not match any required pair of keys, such as F1 and F11 key presses, or F2 and F12 key presses) were recorded separately. Mean RTs were calculated for correct responses only.

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

103

5.1.2. Results and discussion The average percentage of errors of commission was 20.4%. No errors of omission or inconsistent responses were observed. To focus on the comparison between LVF/RH and RVF/LH which is most relevant for present purposes, data from the BVF condition were not entered into the statistical analysis. However, the results from the BVF condition can be seen in Fig. 1. For both mean correct RTs and the percentage of errors of commission, analyses of variance (ANOVAs) with repeated measures on the factors visual field (LVF, RVF) and word familiarity (i.e., word vs pseudoword) and a between-subjects factor gender were performed. The factor gender yielded no significant differences, neither in the main effect nor in any interaction, and is therefore not elaborated on further. The performance data are summarized in Fig. 1. Reaction times. The ANOVA revealed a significant main effect of word familiarity, F ð1; 18Þ ¼ 16:7; p < :001, reflecting faster RTs to words (M ¼ 802 ms) than pseudowords (M ¼ 878 ms). There was also a significant main effect of visual field, F ð1; 18Þ ¼ 7:3, p < :05, reflecting an overall advantage of 35 ms for RVF targets. Finally, there was a tendency for a larger RVF/LH advantage for words than pseudowords, although the interaction between visual fields and word familiarity failed to reach significance, F ð1; 18Þ ¼ 2:2; p ¼ :15. Separate ANOVAs for words and pseudowords suggested a clear RVF advantage of 51 ms for words, F ð1; 18Þ ¼ 12:9; p < :01, whereas the RVF advantage of 18 ms for pseudowords was not significant, F ð1; 18Þ < 1.

Fig. 1. Performance in word recognition in Experiment 1 for words and pseudowords presented in the left and right visual fields, or in both visual fields simultaneously. Top: Reaction times. Bottom: Error rates.

104

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

Error rates. The ANOVA of the error rates revealed a main effect of word familiarity, F ð1; 18Þ ¼ 5:1, p < :05, a main effect of visual field, F ð1; 18Þ ¼ 19:6, p < :001, and an interaction between visual fields and word familiarity, F ð1; 18Þ ¼ 5:9, p < :05. Again, the RVF advantage for words was significant, F ð1; 18Þ ¼ 16:3, p < :001, M ¼ 34:5% vs 18.8% for LVF and RVF trials, respectively. The RVF advantage for pseudowords just failed to reach significance, F ð1; 18Þ ¼ 4:0, p < :06, M ¼ 20:2% vs 15.4% for LVF and RVF trials, respectively. No other effects approached significance. In sum, Experiment 1 yielded the expected pattern of hemispheric asymmetries in a lexical decision task. A clear RVF advantage was seen for common nouns, both in RTs and error rates. In contrast, hemispheric asymmetries were largely absent for pseudowords, suggesting that the observed asymmetries depend on the access to preexisting representations that are only available for words.5 5.2. Experiment 2 Experiment 2 tested whether the RVF/LH advantage would be attenuated or reversed in a task that involves the recognition of famous personal names. All famous names were carefully selected on the basis of a rating study, in line with the suggestion by V&O, to ensure that participants were indeed highly familiar with these stimuli. 5.2.1. Method Participants. The same 20 participants who had contributed data to Experiment 1 also served as participants in Experiment 2. To control for possible systematic effects of the order of experiments, half of the participants performed Experiment 1 followed by Experiment 2, with the order reversed in the other half of the participants. Stimuli and apparatus. The famous names were 32 names of well-known British or American celebrities, selected so that both first name and surname were between 3 and 8 letters in length. A group of 11 independent raters (mostly undergraduate students at the University of Glasgow, like the experimental participants), rated every name out of an original list of 611 famous names according to a 3-point scale from 0 to 2 (0 ¼ unfamiliar, 1 ¼ moderately familiar, 2 ¼ highly familiar). The 32 celebrity names selected for Experiment 2 were those with the highest familiarity ratings (M ¼ 1:99) while fulfilling the length restrictions specified above (see Appendix B).6 Thirty-two unfamiliar names were also generated from the London residential telephone directory (No. 620 British Telecom London Residential, January 2000), with the restriction that unfamiliar names were matched to famous names with respect to number of syllables and letters, ensuring that a familiarity decision could not 5 Inspection of Fig. 1 highlights an additional small point. Performance in the bilateral condition was generally very similar to performance in the better one of the unilateral conditions (i.e., the RVF). This is at least an indirect evidence that the string of X-characters used in the unilateral conditions did not provide ‘‘visual competitive noise’’ that interfered with the task performance (a possible concern by V&O). Had the string of X-characters in the contralateral hemifield interfered strongly with performance, one would have expected a clearly superior performance in the bilateral condition relative to both unilateral conditions. However, there was not much of an improvement in performance in the bilateral condition, relative to the better one of the unilateral conditions (typically the RVF/LH) and a similar result was obtained with personal names in Experiment 2 (see Fig. 2). Thus, the string of X-characters in the contralateral hemifield did not appear to interfere strongly with the performance. 6 Two of these names were actually stage characters which proved to be even more familiar than the respective actors’ names.

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

105

be taken on the basis of superficial characteristics. Stimuli were displayed in white on a dark gray background with the IBM8BIT font at font size 18, with a letter height of 1.1 cm. The stimulus width was between 2.7 and 6.2 cm, depending on the name length, corresponding to a visual angle of between 1.5° and 3.5° at a viewing distance of 1 m. The stimuli were presented at a lateral eccentricity of 5.0 cm (measured from fixation point to stimulus center), corresponding to a visual angle of 2.9°. The first name was presented centered above the surname with a vertical distance of 0.8 cm so that the names encompassed a height of 3.0 cm in total. Procedure. Participants were informed that they would be shown famous and unfamiliar names and that they should perform one of the two different key press responses depending on name familiarity. Both speed and accuracy were emphasized. For famous names, they should press the F1 and F12 keys of a computer keyboard with both the middle fingers. For unfamiliar names, a simultaneous key press of the F2 and F11 keys with both index fingers was requested. For practical reasons the keyboard was turned by 180°. Famous and unfamiliar names were shown either in the LVF or the RVF for 150 ms. At the screen location contralateral to this target stimulus, two strings of Xcharacters (one above the other, to match the spatial characteristics of a pair of first name and surname) were shown simultaneously, with the average number of characters, corresponding to the average number of letters in the target stimuli. All other aspects of the procedure (structure of trials, feedback, assignment of stimuli to visual fields, number of trials, computation of the dependent variables, practice trials, breaks) were analogous to Experiment 1. 5.2.2. Results and discussion The average percentage of errors of commission was 18.1%, with no errors of omission or inconsistent responses. As in Experiment 1, the BVF condition was not included in the statistical analysis to focus on the comparison between LVF and RVF that is most relevant for present purposes. However, performance in all conditions is depicted in Fig. 2. For both mean correct RTs and the percentage of errors of commission, ANOVAs were performed with repeated measures on two two-level factors visual field (LVF, RVF) and name familiarity (famous vs unfamiliar), with a between-subjects factor sex. The factor sex yielded no significant differences in any of the analyses, neither in a main effect nor in any interaction, and is therefore not elaborated on further. Reaction times. The ANOVA revealed a main effect of name familiarity, F ð1; 18Þ ¼ 19:1, p < :001, reflecting faster RTs to famous names (M ¼ 807 ms) than unfamiliar names (M ¼ 929 ms). There was also a main effect of visual field, F ð1; 18Þ ¼ 20:7, p < :001, reflecting an overall advantage of 56 ms for RVF targets and an interaction between visual fields and name familiarity, F ð1; 18Þ ¼ 8:4, p < :01. Whereas there was a large RVF advantage of 108 ms for famous names, F ð1; 18Þ ¼ 31:3, p < :001, asymmetries for unfamiliar names were absent, Mdiff ¼ 3 ms, F ð1; 18Þ ¼ 1:1, p > :20. Error rates. The ANOVA of the error rates revealed a main effect of name familiarity, F ð1; 18Þ ¼ 6:4, p < :05, reflecting slightly more errors to famous names ðM ¼ 20:7%Þ than unfamiliar names ðM ¼ 15:4%Þ. There was also a main effect of visual field, F ð1; 18Þ ¼ 21:9, p < :001, as well as an interaction between visual fields and name familiarity, F ð1; 18Þ ¼ 28:0, p < :001. The RVF advantage for famous names was again significant, F ð1; 18Þ ¼ 31:8, p < :001, M ¼ 36:6% and 15.2% for LVF vs RVF trials, respectively. In contrast, hemispheric asymmetries for unfamiliar names were absent, F ð1; 18Þ < 1, p > :20, M ¼ 14:3% and 16.3% for LVF vs RVF trials, respectively.

106

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

Fig. 2. Performance in name recognition in Experiment 2 for famous and unfamiliar names presented in the left and right visual fields, or in both visual fields simultaneously. Top: Reaction times. Bottom: Error rates.

This experiment confirms a strong RVF/LH advantage for the recognition of personal names in a name familiarity task, replicating our previous findings (Schweinberger et al., 2002). Because V&O had raised the concern that our previous findings of an LH advantage in name recognition may relate to an insufficient familiarity of names, it is important that a rigorous selection of only highly familiar famous names in the present experiments rules out this explanation. Replicating the findings by SLMK, the RVF/LH advantage was specific to famous names and did not show up for unfamiliar names. This finding held both for RTs and error rates and suggests that the LH superiority holds only for personal names with an existing memory representation. Two small but interesting additions to the previous research by SLMK are that we replicated their findings with stimuli from different languages (i.e., English vs German) and that subject gender is unlikely to account for differences in hemispheric asymmetries for name recognition.

6. Conclusion It has been controversial whether or not the RH makes a distinct contribution to the recognition of famous personal names. With a divided visual field technique, the present experiments demonstrated a strong and significant LH superiority in name

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

107

recognition, confirming earlier results by SLMK. The present results differ from data presented by O&V (specifically, their Experiments 1 and 6) and V&O, although they appear to be in line with other data presented by these authors (Experiments 2 and 3 by O&V). A strong role of the left hemisphere in the recognition of famous personal names is in line with a number of studies using functional imaging and ERPs (Dehaene, 1995; Gorno-Tempini et al., 1998; Proverbio et al., 2001; Sergent et al., 1994) and is also consistent with research in brain-injured patients (Newcombe, DeHaan, Ross, & Young, 1989; Schweinberger, 1995; Semenza et al., 1995). Although evidence is growing that personal names are represented by brain systems that differ from those representing common nouns, there is no clear evidence for a distinct role of the RH in the recognition of names. A comparison of the present experiments indicates that we were successful in matching the overall task difficulty for the word and name recognition tasks. When comparing the responses to words and famous names across experiments, it turns out that RTs to names were just 5 ms slower than RTs to words, a nonsignificant difference (M ¼ 802 ms and 807 ms, F ð1; 18Þ ¼ 1:1, p > :10). Error rates to names were slightly lower than to words, a difference which was again not significant (M ¼ 23:6% and 20.7%, F ð1; 18Þ < 1). This comparison also suggests that, if anything, the asymmetry in favor of the LH was even stronger for names compared with words. The interaction between experiment and visual field was significant in RTs, F ð1; 18Þ ¼ 8:7, p < :01. Thus, the 108 ms advantage for the RVF/LH for famous personal names was significantly larger than the 51 ms advantage for the RVF/LH for words. Although a similar effect is seen in error rates, the interaction between experiment and visual field was not significant, F ð1; 18Þ ¼ 2:3, p ¼ :15, indicating that the RVF/LH advantage in error rates for famous names was not significantly larger than the RVF/LH advantage in error rates for nouns (Mdiff ¼ 21.4% vs 15.7%, respectively). We wish to make it clear that in this article we challenged the hypothesis that the RH has a special role in the processing of famous names. We did not make any claims with respect to the validity of Diana VanLancker’s theory that the human RH has a special role in the processing of personally relevant aspects of an individual’s world (VanLancker, 1991). This concept may have merits in the interpretation of functional differences between the cerebral hemispheres. However, it appears to us that if one wishes to evaluate the RH’s role in processing personally relevant stimuli, famous names are not an ideal material. In VanLancker’s own words, personal relevance is not primarily defined by frequency of occurrence of an item, but ‘‘. . .personal relevance requires a relationship, which I shall sometimes call valence, which is defined as the capacity of something to unite, react, or interact with something else. . .’’ (VanLancker, 1991, p. 65). It appears that such a relationship is not typically met by publicly familiar people (famous names), but would require personal familiarity. In this respect, it is important to note that recent research has started to consider the more affective or emotional aspects of person recognition (Breen, Caine, & Coltheart, 2000; Ellis & Lewis, 2001). Also, a recent fMRI study has suggested a role for the retrosplenial cortex bilaterally as a neural correlate of these more personal, affective or emotional aspects of person familiarity (Shah et al., 2001)—although this study used faces and voices, rather than names, as stimuli. In sum, current evidence suggests a strong involvement of the LH in the recognition of famous personal names, but does not confirm a role of the RH. This does not necessarily exclude a contribution of the RH in the recognition of personally familiar names, but even here it remains for future research to see whether such a contribution can be demonstrated.

108

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

Appendix A. Stimuli from Experiment 1: common nouns and pseudowords Common nouns

Frequency/million

Matched pseudoword

CAFE DEBTOR PIPER QUARRY MONARCH LIVER GUINEA RAINBOW MONKEY MORASS NOMAD NOVICE SICKLE TEMPEST PROPANE PANCAKE RIPCORD WAFER WAGON SIGNAL SILAGE SNORKEL TANKER TENDON TAVERN TRUMPET UNIFORM URINE VANDALS BLAZER WEAPON YEOMAN

6 1 2 8 10 7 3 4 2 3 1 4 1 1 3 1 1 1 10 36 12 2 1 2 5 6 7 7 7 4 24 6

CEFA TODREB PREPI RYQUAR ARMCHON VILER NIGUEA WAINBOR MYKEON RASMOS DOMAN CONVIE LEKSIC PETMEST RENAPOP CAKPANE DRIPCOR FERWA GOWAN NALGIS AGELIS KELRONS ERKTAN NEDTON VETARN PUTTREM FORNUMI RUNIE DALVANS BRELAZ WAPNOE MAYONE

Appendix B. Stimuli from Experiment 2: familiar and unfamiliar personal names Familiar names

Familiarity rating

Unfamiliar names

ALAN SHEARER BASIL FAWLTY KATE WINSLET MINNIE DRIVER CLIFF RICHARD EDWINA CURRIE JANET JACKSON ALLY MCCOIST DAMON ALBARN

2.00 2.00 2.00 2.00 2.00 1.91 2.00 2.00 2.00

SIMON HARDY MATTHEW DICKENS PERCY EATON ELAINE ELLIOT IAN COLLIER MARTHA GODDARD DEBBIE HARMAN JEMMA HILLER WALTER JACKSON

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

109

Appendix B (continued) Familiar names

Familiarity rating

Unfamiliar names

ESTHER RANTZEN JENNIFER LOPEZ AUSTIN POWERS JEREMY BEADLE JARVIS COCKER KEVIN KEEGAN LENNY HENRY PHIL COLLINS ULRIKA JONSSON TINA TURNER NEIL KINNOCK DAVID BOWIE UMA THURMAN WHITNEY HOUSTON HARRY ENFIELD GERRY ADAMS JOHN MAJOR QUEEN MOTHER VICTOR MELDREW BOB MARLEY BRYAN ADAMS ELVIS PRESLEY TERRY WOGAN

2.00 2.00 2.00 2.00 2.00 2.00 2.00 2.00 2.00 2.00 2.00 1.91 2.00 1.91 2.00 2.00 1.82 2.00 2.00 2.00 2.00 2.00 2.00

ANNE LIDDLE TRACEY MICHAEL NORMAN MORRIS DAVID NEWMAN ETHEL PALMER SIMON PEYTON CAROL ROBERTS MAURICE SIPSER JOAN SIMONS KAREN THOMSON RONALD TAYLOR LEONARD VINCENT SUSAN WALDING SHEILA BENTLEY GRAHAM WEBSTER KIM YOUNGER IAN LAWFORD EDITH GIBSON ALAN WAKEMAN CHRIS SUMNER FRANCIS ADAMSON DAVID STINSON ALBERT WALTER

References Anstis, S. M. (1974). A chart demonstrating variations in acuity with retinal position. Vision Research, 14, 589–592. Breen, N., Caine, D., & Coltheart, M. (2000). Models of face recognition and delusional misidentification: a critical review. Cognitive Neuropsychology, 17, 55–71. Damasio, H., Grabowski, T. J., Tranel, D., Hichwa, R. D., & Damasio, A. R. (1996). A neural basis for lexical retrieval. Nature, 380, 499–505. Dehaene, S. (1995). Electrophysiological evidence for category-specific word processing in the normal human brain. NeuroReport, 6, 2153–2157. Ellis, H. D., & Lewis, M. B. (2001). Capgras delusion: a window on face recognition. Trends in Cognitive Sciences, 5, 149–156. Gazzaniga, M. S. (1983). Right hemisphere language following brain bisection. A 20-year perspective. American Psychologist, 38, 525–537. Gorno-Tempini, M. L., Price, C. J., Josephs, O., Vandenberghe, R., Cappa, S. F., Kapur, N., & Frackowiak, R. S. J. (1998). The neural systems sustaining face and proper-name processing. Brain, 121, 2103–2118. Grabowski, T. J., Damasio, H., Tranel, D., Ponto, L. L. B., Hichwa, R. D., & Damasio, A. R. (2001). A role for left temporal pole in the retrieval of words for unique entities. Human Brain Mapping, 13, 199–212. Hollis, J., & Valentine, T. (2001). Proper-name processing: Are proper names pure referencing expressions? Journal of Experimental Psychology: Learning Memory, and Cognition, 27, 99–116. Johansson, S. (1989). Frequency analysis of English vocabulary and grammar. Oxford: Clarendon Press. McGlone, J. (1980). Sex differences in human brain asymmetry: a critical survey. The Behavioral and Brain Sciences., 3, 215–263. Mohr, B., Pulverm€ uller, F., & Zaidel, E. (1994). Lexical decision after left, right, and bilateral presentation of content words function words, and non-words: evidence for interhemispheric interaction. Neuropsychologia, 32, 105–124.

110

S.R. Schweinberger et al. / Brain and Language 82 (2002) 95–110

Nass, R. D., & Gazzaniga, M. S. (1987). Cerebral lateralization and specialization in human central nervous system. In V. B. Mountcastle, F. Plum, & S. R. Geiger (Eds.), Handbook of physiology: Section 1: The nervous system. Vol. V: Higher functions of the brain, Part 2 (pp. 701–762). Bethesda: American Physiological Society. Newcombe, F., DeHaan, E. H. F., Ross, J., & Young, A. (1989). Face processing laterality and contrast sensitivity. Neuropsychologia, 27, 523–538. Ohnesorge, C., & VanLancker, D. (1999). Cerebral lateralization of common and proper nouns: evidence from normal processing. Brain Language, 69, 384–387. Ohnesorge, C., & VanLancker, D. (2001). Cerebral laterality for famous proper nouns: visual recognition by normal subjects. Brain Language, 77, 135–165. Oldfield, R. C. (1971). The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia, 9, 97–113. Proverbio, A. M., Lilli, S., Semenza, C., & Zani, A. (2001). ERP indexes of functional differences in brain activation during proper and common names retrieval. Neuropsychologia, 39, 815–827. Ratcliff, R. (1985). Theoretical interpretations of the speed and accuracy of positive and negative responses. Psychological Review, 92, 212–225. Schweinberger, S. R. (1995). Personal name recognition and associative priming in patients with unilateral brain damage. Brain & Cognition, 29, 23–35. Schweinberger, S. R., Landgrebe, A., Mohr, B., & Kaufmann, J. M. (2002). Personal names and the human right hemisphere: An illusory link? Brain Language, 80, 111–120. Schweinberger, S. R., Sommer, W., & Stiller, R. M. (1994). Event-related potentials and models of performance asymmetries in face and word recognition. Neuropsychologia, 32, 175–191. Semenza, C., Mondini, S., & Zettin, M. (1995). The anatomical basis of proper name processing. A critical review. Neurocase, 1, 183–188. Sergent, J., MacDonald, B., & Zuck, E. (1994). Structural and functional organization of knowledge about faces and proper names: a positron emission tomography study. In C. Umilta, & M. Moscovitch (Eds.), Attention and Performance XV (pp. 203–228). Cambridge, London: MIT Press. Shah, N. J., Marshall, J. C., Zafiris, O., Schwab, A., Zilles, K., Markowitsch, H. J., & Fink, G. R. (2001). The neural correlates of person familiarity. A functional magnetic resonance imaging study with clinical implications. Brain, 124, 804–815. VanLancker, D. (1991). Personal relevance and the human right hemisphere. Brain & Cognition, 17, 64–92. VanLancker, D., & Klein, K. (1990). Preserved recognition of familiar personal names in global aphasia. Brain & Language, 39, 511–529. VanLancker, D., Klein, K., Hanson, W., Lanto, A., & Metter, E. J. (1991). Preferential representation of personal names in the right hemisphere. Conference on Clinical Aphasiology, 20, 181–190. VanLancker, D., & Ohnesorge, C. (2002). Personally familiar proper names are relatively successfully processed in the human right hemisphere or the missing link. Brain & Language, 80, 121–129. Warrington, E. K., & Clegg, F. (1993). Selective preservation of place names in an aphasic patient: a short report. Memory, 1, 281–288. Zaidel, E. (1983). Language in the right hemisphere, convergent perspectives. American Psychologist, 38, 542–549.