Memory for environmental sounds in sighted, congenitally blind and late blind adults: evidence for cross-modal compensation

International Journal of Psychophysiology 50 (2003) 27–39

Memory for environmental sounds in sighted, congenitally blind and late blind adults: evidence for cross-modal compensation ¨ ¨ Brigitte Roder*, Frank Rosler Department of Psychology, Philipps-University Marburg, Gutenbergstrasse 18, 35032 Marburg, Germany Received 8 August 2002; accepted 10 January 2003

Abstract Several recent reports suggest compensatory performance changes in blind individuals. It has, however, been argued that the lack of visual input leads to impoverished semantic networks resulting in the use of data-driven rather than conceptual encoding strategies on memory tasks. To test this hypothesis, congenitally blind and sighted participants encoded environmental sounds either physically or semantically. In the recognition phase, both conceptually as well as physically distinct and physically distinct but conceptually highly related lures were intermixed with the environmental sounds encountered during study. Participants indicated whether or not they had heard a sound in the study phase. Congenitally blind adults showed elevated memory both after physical and semantic encoding. After physical encoding blind participants had lower false memory rates than sighted participants, whereas the false memory rates of sighted and blind participants did not differ after semantic encoding. In order to address the question if compensatory changes in memory skills are restricted to critical periods during early childhood, late blind adults were tested with the same paradigm. When matched for age, they showed similarly high memory scores as the congenitally blind. These results demonstrate compensatory performance changes in long-term memory functions due to the loss of a sensory system and provide evidence for high adaptive capabilities of the human cognitive system. 䊚 2003 Elsevier Science B.V. All rights reserved. Keywords: Cross-modal compensation; Long-term memory; Blindness; False memories; Level-of-processing

1. Introduction It is commonly assumed that blind people compensate for visual information by an enhanced use of input provided by their intact sensory systems, although empirical data on this issue are very inconsistent. In part, this inconsistency may be due to the employed criteria for the recruitment of *Corresponding author. Tel.: q49-6421-282-3723; fax: q 49-6421-282-8948. ¨ E-mail address: [email protected] (B. Roder).

blind participants in different studies (e.g. varying degrees and duration of blindness, etiology of blindness, age, the type or lack of a matching control group), the use of non-identical tasks for sighed and blind participants, different pre-experimental experience across groups etc. (for a detailed ¨ discussion, see e.g. Millar, 1982; Roder and Neville, 2003; Thinus-Blanc and Gaunet, 1997). Recently, an increasing number of studies have reported more efficient perceptual processing (e.g. shorter reaction times) in blind compared to sight-

0167-8760/03/$ - see front matter 䊚 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0167-8760(03)00122-3

28

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

ed people, both in auditory and tactile discrimi¨ nation tasks (Kujala et al., 1997; Roder et al., 1996). Moreover, neurophysiological recordings have revealed similar neural changes in the blind as have been observed in populations with a specific history of perceptual experience (e.g. musicians). For example, both musicians (Pantev ¨ et al., 1998) and blind adults (Roder et al., 1999a), show an enhanced excitability of neural systems important for auditory processing. It has, therefore, been argued that compensatory performance changes in the blind may, at least partly, be due to perceptual learning. In addition, some brain imaging studies have revealed posteriorly extended brain activation patterns in blind compared to sighted individuals, suggesting that a reorganization of multisensory brain areas and of brain regions primarily associated with visual processing in sighted individuals may contribute to improved perceptual skills in the blind (Cohen et al., 1997; ¨ Kujala et al., 1995; Roder et al., 1999b, 2002, 2003). In addition to investigating compensatory changes, studying blind individuals can also help to identify the unique contribution of visual input for particular perceptual-cognitive functions, e.g. spatial cognition (Hollins, 1989; Thinus-Blanc and Gaunet, 1997) or language acquisition (Andersen et al., 1993; Mills, 1988), It has, for example, been tested if blind people (who have no visual imagery) show less gain from imagery strategies, which are known to improve memory performance in the sighted. It has been reported that blind individuals remembered items with high imagery scores with a higher probability than items with low imagery scores, just as sighted participants (Jonides et al., 1975; Paivio and Okovita, 1971), unless imagery ratings were exclusively based on visual properties (Marchant and Malloy, 1984). Since blind people rely more heavily on auditory information, it has been argued that they should show superior memory for input delivered through this modality. However, Cobb et al. (1979) did not find any differences in long-term memory for environmental sounds, nor for common tactile ¨ objects, between sighted and blind adults. Ronnberg and Nilsson (1987) suggested that the auditory and visual system differ with respect to the

depth of processing, i.e. that a visually based code is more permanent than an auditory code. In favor of their hypothesis, they did not find compensatory memory performance in their blind sample, but rather report ‘«signs of inferior performance.’ ¨ (Ronnberg and Nilsson, 1987, p. 279). However, ¨ the blind sample of Ronnberg and Nilsson comprised late blind adults with different degrees of visual impairments. If the development of compensatory memory functions is linked to critical periods as has been observed for perceptual (Hubel and Wiesel, 1977) or language functions (Curtiss, 1977; Neville et al., 1997), the lack of compen¨ sation found by Ronnberg and Nilsson (1987) could be explained. Nevertheless, Pring (1988) and Pring et al. (1990) provided evidence for different memory strategies in blind and sighted adolescents. They tested memory for words which had either been read or heard in a group of sighted and blind pupils. Stimuli were either presented or had to be actively generated by the participants themselves. In the sighted control group, generated words were recalled with a higher probability than the words that were either just heard or read, replicating the so-called ‘generation effect’. In contrast, the blind group showed a reversed result pattern. Since, the ‘generation effect’ is thought to be based on the ‘enhanced conceptual distinctiveness’ of self-produced items, the lack of a generation effect in the blind was attributed to an impaired or less well elaborated semantic network, which was assumed to be a consequence of the lack of visual input in the blind. They have to acquire many concepts through language with less or without direct sensory experience. Therefore, Pring (1988) hypothesized that their semantic networks contain more abstract concepts. Furthermore, it was consequently argued that the blind prefer data-driven strategies. That is, they remember items on the basis of (non-visual) sensory features, rather than on the basis of conceptual relations. However, this hypothesis was not confirmed in a word completion instead of a recall task (Pring et al., 1990). Nevertheless, the lack of group differences may be due to the fact that word completion is an implicit memory task, which is only indirectly related to sensory codes, since physical symbols (words presented in Braille)

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

were presented rather than sensory features of the concepts proper. It is known that the development of blind children is delayed by 1–2 years (Warren, 1994). Therefore, the results of Pring et al. with adolescents may not hold true with adults. Accordingly, Kool and Rana (1980) showed that the efficiency of tactual encoding and memory for tactual objects improves with age. In their study, an initial performance inferiority of the blind during childhood turned into a superiority during adolescence. The present study was designed to test the efficiency of different encoding strategies for auditory recognition memory in blind compared to sighted adults. It is known from the level-ofprocessing approach (Craik and Tulving, 1975) that semantic encoding results in higher recognition scores than physical encoding. It has been assumed that conceptual processing increases the durability of memory traces. If blind people in fact make increased use of the sensory features of the to-be-remembered items, they should gain particularly from physical encoding instructions. Furthermore, if semantic networks are impoverished in the blind, they should gain less than sighted individuals from semantic encoding strategies. In the present study, a typical level-of-processing paradigm was used in which participants either encoded environmental sounds physically or semantically. In the recognition phase, the previously heard ‘old’ sounds were presented intermixed both with new sounds that were conceptually and physically distinct from the previously encountered sounds and new sounds that were physically distinct but belonged to the same concept (i.e. were named the same, for example the barking of two different dogs) as one of the environmental sounds heard in the study phase. It is well known from the ‘false-memory’ literature that false alarm rates increase with semantic relatedness between old items and distracter items (Roediger and McDermott, 1995; Schacter, 1997). Therefore, it was hypothesized that semantic encoding should increase false memory effects while physical encoding should reduce false memory effects, particularly in congenitally blind participants.

29

Studies on neural plasticity have reported extensive reorganization in both the developing as well as adult system of animals (e.g. Kaas, 2000; Recanzone, 2000) and humans (Elbert et al., 1995; Flor et al., 1995; Pantev et al., 1998). However, it is also widely accepted that the development of some perceptual-cognitive functions is linked to critical periods in early childhood, e.g. including the development of binocular functions (e.g. Hubel and Wiesel, 1977) or the acquisition of grammatical aspects of language (Neville et al., 1997; Weber-Fox and Neville, 1996). It is interesting to note that sensory map changes in musicians (Elbert et al., 1995; Pantev et al., 1998) are larger the earlier the perceptual training had started. It could be hypothesized that the development of different perceptual-cognitive functions show different degrees of vulnerability to altered environmental conditions. While the establishment of some functions seems to depend completely on adequate input during restricted time periods in early childhood, there may only be some gain from the higher plasticity capacity during early development for other functions (although they are not completely dependent on early experience). To test if cross-modal compensatory performance changes in auditory memory for environmental sounds can evolve in adulthood, we tested a group of late blind participants. If the compensatory capacity of auditory memory functions is lower in adulthood than in early childhood, both higher memory performance for the late blind than for the sighted groups as well as lower memory scores than for the congenitally blind groups would be predicted. If, however, compensatory plasticity of auditory memory is restricted to limited time periods during early childhood, no difference between late blind and sighted individuals would be expected. 2. Materials and methods 2.1. Participants A group of 24 sighted university students (mean age 21.5 years, range 19–25 years, 16 female, 3 left-handed), 20 congenitally blind adults (mean age 21.4, range 18–32 years, 11 females, 2 left-

30

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

handed, 3 ambidextrous) and 20 late blind adults (mean age 35.3 years, range 21–53 years, 5 female, 2 left-handed, 1 ambidextrous) participated. The congenitally blind participants were either university (Ns7) or high school students (Ns 12); and one was a computer administrator. They were either totally blind (Ns10) or had some rudimentary sensitivity for light. Blindness was congenital and due to peripheral defects including retinotopia of prematurity (Ns9), congenital glaucoma (Ns2), retina detachment (Ns3), optic nerve atrophy (Ns2), blurring of the aqueous chamber (Ns1), inherited defects (Ns1), or unknown eye defect (Ns2). Eight of the late blind participants were totally blind. The remaining had some sensitivity for brightness changes. Blindness had prevailed for at least 5 years (means14.5 years, range 5–32 years) and was due to peripheral defects caused by retina detachment (Ns4), retinitis pigmentosa (Ns6), glaucoma (Ns2) optic nerve atrophy (Ns1), cataract (Ns1) or an accident (Ns6). Although half of the late blind participants had had visual impairments before the age of 12, they had all been able to read print. All participants reported normal hearing. They were either monetarily compensated or received course credits. 2.2. Materials A total of 226 environmental sounds (duration varied between 400 and 800 ms) from one of the following categories were used: animals (e.g. lion, bee), public life (traffic (e.g. motor cycle, horn)), tools (e.g. hammer, saw), army (e.g. machinegun, shot), house (e.g. door, glass), human (e.g. kiss, weeping), hobby (e.g. billiard, camera), instrument (flute, drum). The 8 practice sounds and 118 experimental sounds were selected from a pool of 320 environmental sounds on the basis of a pilot study, in which 11 sighted university students had assigned a name to each sound. Only sound stimuli with a concept agreement of equal or higher than 0.5 were included (i.e. half of the participants assigned either the same or a semantically closely related name to the sound). The stimuli were presented with a loudspeaker, located 0.5 m in

front of a participant (sound level at the participant’s ear was approximately 65 dB(A)). The experiment was computer controlled using the program ERTS (Experimental Run Time System, Berisoft Corporation, Frankfurt, Germany). 2.3. Procedure In the study phase, participants heard 59 environmental sounds. Half of the participants of each group had to name the sounds (semantic encoding) and the other half had to rate the sounds as ‘harsh’ or ‘soft’ using a five-level rating scale (physical encoding). An experimenter entered the responses of the participants into the computer. After the study phase, participants were engaged in a distraction task in order to eliminate any possible contribution of short-term memory to recognition performance: six sinusoidal tone pairs were presented in random order (1000 Hz vs. 950 Hz, 1000 Hz vs. 1050 Hz, 900 Hz vs. 950 Hz and in reversed pairings; duration of each tone: 100 ms, stimulus onset asynchrony: 500 ms) and participants had to decide which of the two tones was higher in pitch. In the consecutive recognition phase participants heard the 59 sounds of the study phase randomly mixed with 59 new sounds. They had to decide whether the identical sound had or had not been presented in the study phase. One second after the oldynew response, a tone (400 Hz, 100 ms) was presented, asking the participant to rate the confidence of her decision. Participants responded with the left and right mouse button using their dominant hand. Response-category and button assignment was counterbalanced across participants. While 70% of the new sounds belonged to a concept which had not been used in the study phase, the remaining 30% of the new sounds were a second example of a concept that had occurred in the study phase. That is, in the pilot study they had been named the same as one sound in the study phase but were physically distinct from them. Hereafter, they will be called pseudo-old items. Since, participants were instructed to decide whether or not they had heard the identical sound during study, a new response was required for both conceptually distinct and conceptually non-distinct new items. Accuracy was stressed, that is partici-

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

31

pants had 10 s to indicate their old-new decision and another 20 s to rate their confidence. No feedback was given. Stimuli were counterbalanced across studied and non-studied conditions across participants, such that the study and distracter items for half of the participants were distracter and study items, respectively, for the other half of the participants. In order to familiarize participants with the procedure, two study-recognition loops were run with four (study phase)yeight (recognition phase) additional items before the experiment proper was started. Sighted participants were blindfolded throughout the session. The experiment lasted for approximately 45 min. 2.4. Data analysis First, memory performance was parameterized by combining hit and false memory rates to the sensitivity value d9 (signal detection theory; Green and Swets, 1966) which was submitted to an ANOVA with between-participant factors Visualstatus (sighted, congenitally blind, late blind) and Study-mode (semantic vs. physical encoding). Pseudo-old items were not considered for this analysis. The same ANOVA was run with the response criteria (BETA of the signal detection theory) in order to assess possibly different strategies in blind and sighted participants. ‘False memory’ was assessed by subtracting false alarm rates for distinct-new items from false alarm rates to pseudo-old items. These ‘False memory rates’ were tested against zero, with a significant deviation from zero indicating that false alarms were higher for pseudo-old items than distinct-new items. They were then submitted to an ANOVA with factors of Visual-status (sighted, congenitally blind, late blind) and Study-mode (semantic vs. physical encoding). In order to assess a possible contribution of differences in age and duration of blindness, late blind participants were divided into (a) a young and old group and (b) in a long-duration and short-duration blind group. Memory performance (d9) and false memory rates were compared between the two late blind groups, respectively, and also between these late blind and the congenitally blind groups.

Fig. 1. Mean d9 (with standard error bars) for sighted (left), congenitally blind (middle) and late blind (right) participants as a function of study mode.

Finally, percentage scores of the correct discrimination responses in the distraction task (tone discrimination task) were calculated and submitted to an ANOVA with factors Visual-status (sighted, congenitally blind, late blind) and Study-mode (semantic vs. physical encoding). Statistical analyses were run with the SAS statistics software package. The GLM procedure was used, taking into account the different group sizes. 3. Results 3.1. General differences between sighted and blind groups 3.1.1. d9 The overall ANOVA with between factors Visual-status (sighted, congenitally blind, late blind) and Study-mode (physical vs. semantic) revealed a significant main effect of Visual-status (F(2, 58)s5.07, Ps0.0093) (Fig. 1). Post-hoc Scheffetests confirmed that memory scores were significantly higher for the congenitally blind (M(CB)s 2.21, S.E.s0.14) compared to the sighted (M(Sigh.)s1.69, S.E.s0.12) (P-0.05), whereas

32

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

Fig. 2. Mean false memory rates (with standard error bars) for sighted (left), congenitally blind (middle) and late blind (right) participants as a function of study mode.

those from the late blind (M(LB)s1.95, S.E.s 0.16) did not significantly differ from either the sighted or the congenitally blind. Semantic encoding resulted in better recognition performance than physical encoding (F(1, 58)s26.17, Ps0.0001; (M(phy.)s1.59, S.E.s0.10; M(sem)s2.28, S.E.s0.10), irrespectively of Visual-status (Visual-status X Study-mode interaction: P)0.66). 3.1.2. Beta The response criteria did not differ between sighted, congenitally blind, and late blind groups (Visual-status: F(2, 58)s0.25, Ps0.7818; M(Sigh.)s1.21, S.E.s0.11; M(CB)s1.14, S.E.s0.07; M(LB)s1.22, S.E.s0.08). Physical encoding participants were generally more conservative than semantic encoding participants (F(l, 58)s11.39, Ps0.0013; M(phy.)s1.35, S.E.s 0.08; M(sem.)s1.03, S.E.s0.05) irrespectively of Visual-status (Visual-status X Study-mode: P) 0.39). 3.1.3. False memory scores In all groups, false alarm rates were significantly higher for pseudo-old items than for distinct-new items (P-0.0001).

The overall ANOVA revealed a marginally significant main effect of Visual-status (F(2, 58)s 2.84, Ps0.0666), indicating a trend towards lower false memory rates in the congenitally blind than the remaining groups (Fig. 2). Physical encoding resulted in lower false-memory rates than semantic encoding (F(1, 58)s7.40, Ps0.0086). Because of our specific predictions and the low sensitivity of a complete between-participants design for Treatment X Group interactions (Kirk, 1982), specific contrasts were calculated for each level of Visual-status, although the Visual-status X Studymode interaction did not reach significance (F(2, 58)s2.00, Ps0.1440).The Study-mode manipulation was significant for the congenitally (P0.05) and the late blind (P-0.01) (Fig. 2). Pair-wise comparisons between the three physical encoding and the three semantic encoding groups revealed only one significant result: the physical encoding congenitally blind group had lower false memory rates than the physical encoding sighted group1,2 (P-0.05). 3.1.4. Distraction task: error rates The three groups did not differ in their performance in the tone discrimination task used for distraction (M(Sigh)-0.60, S.E.s0.04; M(CB)s 0.76, S.E.s0.04; M(LB)s0.07, S.E.s0.02) (F(2, 58)s0.64, P)0.53). Likewise, study-mode did not affect tone discrimination performance (P) 0.99), suggesting that Visual-status and Studymode differences were not due to a different contribution of short-term memory effects. 1 Due to different gender distributions in the groups possible effects of gender were assessed both within groups (across factor encoding) and for the two encoding conditions across groups. There was no significant gender effect for any of the dependent variables (d9, BETA, False memory rate). 2 The pattern of confidence ratings was the same for the sighted, congenitally and late blind groups. Semantic encoding was associated with higher confidence than physical encoding (P)0.05) and incorrectly as old classified pseudo-old items resulted in lower confidence than correctly as old classified old items (P-0.0001); correctly as new classified distinctnew items were associated with higher confidence than correctly as new classified pseudo-old items (P-0.0001). Incorrectly as old classified distinct-new items received lower confidence ratings than incorrectly as old classified pseudoold items (P-0.0001).

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

3.1.5. Discussion The present finding of superior memory in congenitally blind as compared to age and education matched sighted controls adds to the increasing number of recent studies reporting compensatory changes in different auditory perceptual-cognitive tasks (Kujala et al., 2000; Muchnik ¨ et al., 1991; Roder et al., 1999b, 2000), and ¨ memory functions in particular (Roder et al., 2001). Furthermore, they are incompatible with the hypothesis of less elaborated semantic networks in congenitally blind individuals (Pring, 1988; Pring et al., 1990). In contrast, semantic encoding as compared to physical memorizing strategies improved recognition performance in the congenitally blind in a very similar way as in the sighted and in the late blind. The average memory scores of the late blind were somewhat, although not significantly, higher than those of the sighted and somewhat, although again not significantly, lower than those of the congenitally blind. At first glance, one may conclude that the first of the proposed hypotheses is confirmed—compensatory plasticity for auditory memory is higher in childhood but exists, although to a lesser extent, in adulthood too. The nonsignificant comparisons of the late blind with any other group suggest a high variance among the late blind participants, which may originate in differences in chronological age and duration of total blindness. Unfortunately, duration of blindness and chronological age are highly confounded. The late blind group in this study was relatively large (20 participants) and very heterogeneous with respect to age at blindness onset, duration of blindness, and the presence of visual impairments before turning totally blind. This allowed for their division into a young vs. old late blind group, a long- vs. short-duration-late blind group, and late blind groups with and without visual impairments before turning totally blind. In turn, these subdivisions permitted an assessment of the effects of absolute age, duration of blindness and the presence of early visual impairments, respectively.

33

of the congenitally blind and sighted participants and into a group of participants who were older than the congenitally blind and sighted participants. These were compared to both the congenitally blind and sighted groups. The young late blind group consisted of 8 participants (1 female, mean age: 26.5 years, range 21–31 years, all righthanded), whose blindness had prevailed for an average of 10 years (range 5–14 years). Four young late blind participants had encoded the environmental noises physically, and 4 semantically. The average age of the remaining 12 late blind participants was 41 years (range 33–53 years, 4 female, 2 left-handed, 1 ambidextrous); total or nearly total blindness existed on average for 16 years (range 5–32 years). 3.2.1. d9 First, the young and old late blind participants were compared. The ANOVA revealed a significant age effect (F(1, 16)s11.38, Ps0.0039), indicating higher memory scores in the young than the old late blind groups (Fig. 3a). In both groups, semantic encoding resulted in higher recognition than physical encoding (F(1, 16)s6.00, Ps 0.0262). The young late blind groups were compared with the sighted and congenitally blind groups as well. The overall ANOVA revealed a main effect of Visual-status (F(2, 46)s8.95; Ps 0.0005) and a main effect of Study-mode (F(1, 46)s26.50, Ps0.0001). Conservative post-hoc tests (Scheffe) indicated that, while the congenitally and late blind groups did not differ in their memory performance, both groups had higher memory scores than the sighted groups (P-0.05). 3.2.2. False memory rates While the young and old late blind groups did not differ in their false memory rates (P)0.5), they were lower with physical than with semantic encoding (F(1, 16)s9.18, Ps0.0080) (Fig. 3b). The young late blind group differed neither from the sighted nor from the congenitally blind groups (P)0.2). Therefore, false memory rates do not seem to be affected by age.

3.2. Effects of age

3.3. Effects of blindness duration

Late blind participants were divided into a group of participants whose age was within the age range

To investigate if the duration of blindness affects memory performance, a group of late blind partici-

34

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

Fig. 3. Mean d9 (a) and false memory rates (b) (with standard error bars) for young (left) and old (right) late blind participants as a function of study mode.

pants was composed who had been blind for a similar amount of time as the congenitally blind. Late blind participants who had been totally blind for more than 15 years (long-duration-late blind group) consisted of 7 participants (2 females, mean age: 43 years, range 34–53 years, 1 left-handed). They did not have more than brightness perception for an average of 21.5 years (range 16–32 years). Four had encoded the environmental noises physically and 3 semantically. Average age of the remaining 13 late blind participants was (shortduration-late blind group) 31 years (range 21–44 years, 3 female, 1 left-handed, 1 ambidextrous); total or nearly total blindness existed in these participants for in average 9.5 years (rang 5–14 years). 3.3.1. d9 The comparison of the memory scores of the long-duration vs. short-duration late blind participants failed to reach significance (F(1, 16)s2.44, Ps0.1382) (Fig. 4a). The overall ANOVA comparing long-duration late blind, congenitally blind, and sighted participants revealed a significant Visual-status effect (F(2, 45)s7.31, Ps0.0018). Post-hoc Scheffe tests showed that the congenitally

blind had higher memory scores than the sighted and long-duration blind participants (P-0.05), while the latter two groups did not differ. Semantic encoding resulted in higher recognition scores than physical encoding (F(1, 45)s28.76, Ps0.0001), irrespective of Visual-status. 3.3.2. False memory rates Long- and short-duration late blind groups did not differ in their false memory rates (F(1, 16)s 0.14, P)0.7) (Fig. 4b). Physical encoding resulted in lower false memory scores than semantic encoding (F(1, 16)s8.84, Ps0.0090). There was no significant difference between the long-duration late blind and the congenitally blind (F(2, 45)s 1.70, Ps0.1942). False memory rates were lower in the physical encoding groups than in the semantic encoding groups (F(1, 45)s5.29, Ps0.0261), irrespective of Visual-status. 3.4. Effect of visual impairments before turning totally blind Memory performance did not significantly differ (P)0.89) between those late blind participants with visual impairments prior to age 12 (Ns10;

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

35

Fig. 4. Mean d9 (a) and false memory rates (b) (with standard error bars) for long-duration (left) and short-duration (right) late blind participants as a function of study mode.

aged 25–53; mean 35 years; 1 female; 5–32 years of total or nearly total blindness with an average of 10 years) and those without impairments prior to age 12 (Ns10; aged 21–47; mean 35 years; 4 female 9–18 years of total or nearly total blindness with an average of 16 years). The groups of the participants with visual impairments before age 12 years differed neither from the congenitally blind groups nor from the sighted groups. The same pattern held true for false memory rates.

age, the late blind performed as well as the congenitally blind. The response criteria did not differ between sighted, congenitally blind, and late blind groups. These findings supplement recent findings on cross-modal compensation after sensory deprivation and provide insight into contributing factors.

4. General discussion

Within the auditory modality, a superiority of congenitally blind individuals has been reported ¨ for simple auditory discrimination tasks (Roder et al., 1996, 1999a), as well as more complex tasks such as auditory localization (Lessard et al., 1998; ¨ Roder et al., 1999b) and language processing ¨ (Roder et al., 2000, 2002, 2003). With respect to memory functions, both short-term (e.g. digit and ¨ word span; Hull and Mason, 1995; Roder and Neville, 2003), and long-term memory capacities ¨ (e.g. for voices Bull et al., 1983; Roder and ¨ Neville, 2003 or verbal material; Roder et al., 2001) were reported to be enhanced in the blind. Nevertheless, not all research has shown superior memory in blind people (Miller, 1992), particular-

The present study focussed on cross-modal compensation in auditory long-term memory functions after congenital and late blindness. It was tested if blind adults can gain from semantic (‘deep’) as compared to physical (‘shallow’) encoding in a recognition task for environmental sounds in a similar fashion as sighted people. Sighted, congenitally, and late blind participants had higher recognition scores after semantic than physical encoding. However, the congenitally blind outperformed and also showed lower false alarm rates than the sighted for conceptually similar items after physical encoding. When matched for

4.1. Cross-modal compensation in congenitally blind adults

36

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

ly not for environmental sounds (Cobb et al., 1979). However, in that study, detailed information about the degree and etiology of blindness and educational background were not reported. Moreover, the age of the 9 blind participants varied between 20 and 40 years, whereas all controls were college students. The present findings for the late blind suggest that the missing group differences may be due to age differences, which undermined possible advantages of the blind. In general, our results are not consistent with the assumption that semantic networks in the blind are less elaborated and that their concepts are more abstract. In the present investigation, both groups of blind participants were superior to sighted participants under both physical and semantic encoding conditions and, importantly, showed higher recognition scores after semantic than physical encoding. It could be argued that the congenitally blind compensated for the lack of visual input by developing conceptual networks with more acoustical and tactile nodes. Their lower false memory rates compared to those of the sighted when attending to physical stimulus features during study provides evidence for this hypothesis. The question of which mechanisms are responsible for compensatory memory performance in the blind, e.g. whether encoding andyor retrieval are more efficient, can only be answered with additional ¨ neurophysiological measures. Roder et al. (2001) found higher recognition for spoken words in a congenitally blind than an age and education matched sighted group. Moreover, differences between sighted and blind participants in the concurrently recorded event-related brain potentials (ERP) emerged already during study: Only blind participants showed differences in brain activation depending on whether or not words were later recognized. This effect was most pronounced over the left frontal cortex which, based on brain imaging studies (Wagner et al., 1998), suggests an engagement of semantic encoding strategies, known to result in higher recognition scores. In addition, ERP differences between successfully and unsuccessfully recognized words were more pronounced in the congenitally blind, particularly in a late component over the right frontal cortex. This late effect has been associated with post-

retrieval evaluation and retrieval monitoring (e.g. Weyerts et al., 1997). Thus, an improved retrieval monitoring during recognition may have contributed to the elevated memory in the congenitally blind as well. Which factors may account for the different results of our study (no evidence for a semantic processing deficit) and those of Pring et al. (Pring, 1988; Pring et al., 1990) (evidence for a semantic processing deficit)? The most obvious difference between both studies is the mean age of the participants. While the present study tested only adults, the mean age of the participants of Pring et al. was 14 years. Since, it is well known that the development of blind children and adolescents can be delayed up to 2 years (Warren, 1994), blind children or adolescents might in fact be less able than their sighted counterparts to make use of conceptual knowledge to aid memory or might simply have had less experience and knowledge about the employed stimuli.3 Moreover, ‘generation’ (Pring et al.) and ‘naming’ (present study) tasks may target semantic networks in a different way. In sum, the present findings and results from language studies in the blind (Hamann, 1996; ¨ Roder et al., 2000, 2002, 2003) do not support the hypothesis of generally impoverished conceptualsemantic networks in the blind adults. 4.2. Cross-modal compensation in late blind adults The analyses based on the post-hoc grouping of the late blind participants revealed that absolute age rather than duration of blindness seems to determine memory performance in the late blind. Compared with sighted and congenitally blind participants of about the same age, late blind participants were superior to the sighted while they did not differ from the congenitally blind. All late blind participants had been totally or nearly totally blind for at least 5 years. Thus, this time period seems to be sufficient to acquire compensatory memory skills. Since, these adaptations emerge after age 12 years, compensatory memory changes, as measured in the present investigation, are most likely not linked to critical periods in early child3

We thank an anonymous reviewer for this suggestion.

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

hood. This finding is in agreement with the results of Bull et al. (1983), who systematically examined the effects of age at onset and duration of blindness on voice recognition abilities. As in the present investigation, no influence of these factors was found. Rather, the blind as a whole performed better than a sighted control group. The effect of absolute age was not examined by Bull et al., but they had included only adults younger than 42 years in their sample. Similarly, as has been argued for spatial abilities in the blind (Thinus-Blanc and Gaunet, 1997), one could propose that both visual experience before and the higher reliance on auditory information since the onset of blindness results in a double advantage for the late blind adults. If this were true, late blind participants should have shown better memory than congenitally blind participants. Since, this was not the case, the current results do not provide evidence for an additional gain from conceptual knowledge acquired through the visual channel before the onset of blindness. Nevertheless, based on the present data, it is not possible to decide if the same mechanisms mediate compensatory behavior changes in congenitally and late blind individuals. For example, one could argue that auditory compensation in the late blind is only partial and it is their early visual experience that allows them to keep up with the congenitally blind. 4.3. False memory effect In recognition experiments, false memory is particularly high for lures that are semantically associated with items of the study list (Deese, 1969). According to the ‘implicit associative response’ account, this is due to the fact that participants generate or activate concepts related to the presented items during study. If these related items are used as distracters during recognition, the probability to commit a false alarm response is particularly high. Similarly, the ‘pattern separation failure’ account proposes ‘«a high inter-item similarity, resulting in robust memory for ‘gist’ information about perceptual or conceptual features of studied pictures, but poor memory for picture-specific details’ (Schacter et al., 1998, p.

37

297). For example, if previously seen pictures are presented mixed with new pictures, false alarm rates are particularly high for lures that belong to the same category as pictures seen during the study phase (Koutsaal and Schacter, 1997 in Schacter et al., 1998). In the present study, pseudo-old items were selected in a way to guarantee the highest possible associative relation with items presented in the study list. That is, although they were physically distinct, the same names had been assigned to these lures in the pilot study as to previously heard sounds. As predicted (Deese, 1969; Roediger and McDermott, 1995), higher false alarm rates were obtained for pseudo-old items than for conceptually and physically distinct lures (reflected in the significant false memory rates). It has been reported that, while a priori information about the intention of the experiment (implement false memories) does not prevent false recognition (Schacter et al., 1998), additional itemspecific information (e.g. additional line drawings for presented words) reduces false alarms for semantic related items (Israel and Schacter, 1997, in Schacter et al., 1998). Similarly, we had hypothesized that focussing on physical features during study should lead to lower false recognition than focussing on semantic aspects of the study items. However, study mode influenced false memory rates reliably only in the blind. This implies that blind participants were able to improve recognition by particularly attending to acoustic item features during study, which may be due to their elevated capabilities to encode ¨ and use auditory–sensory features (Roder, et al., 1999a). 4.4. Summary and conclusions The present study provides evidence for compensatory memory abilities for auditory stimuli following the loss of visual input. It is shown that auditory long-term memory functions improve their capacities even in adulthood. The results lead to two conclusions. One, they contradict the assumption that blind people are less able to use conceptually based encoding strategies. Two, they imply that blind people are able to make elevated use of perceptual encoding to aid recognition, if

38

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39

they explicitly attend to physical stimulus features during encoding. Future studies should test if the neural mechanisms underlying behavioral improvements are the same in congenital and late blindness, i.e. during early development and in adulthood. Acknowledgments We thank the Study Center for the Blind (Deutsche Blindenstudienanstalt, Marburg) and the German Society for the blind and the visually handicapped in study and occupation (DVBS) for help in recruiting blind participants. This work was supported by grant Ro 1226y4-1 of the German Research foundation (DFG) and the German American Academic council foundation (GAAC). We thank Dr Micah M. Murray and Dipl.-Psych. Tobias Schicke for careful proof reading and Agnieszka Czajkowska, Antje Gubitz, Petros Stathakos, Nicole Sandelbaum and Lars Stetten for their help during data acquisition. References Andersen, E.S., Dunlea, A., Kekelis, L., 1993. The impact of input; language acquisition in the visually impaired. First Lang. 13, 23–49. Bull, R., Rathborn, H., Clifford, B.R., 1983. The voicerecognition accuracy of blind listeners. Perception 12, 223–226. Cobb, N., Lawrence, D.M., Nelson, N.D., 1979. Report on blind subjects’ tactile and auditory recognition for environmental stimuli. Percept. Motor Skills 48, 363–366. Cohen, L.G., Celnik, P., Pascual-Leone, A., et al., 1997. Functional relevance of cross-modal plasticity in blind humans. Nature 389, 180–183. Craik, F.I.M., Tulving, E., 1975. Depth of processing and the retention of words in episodic memory. J. Exp. Psychol. wGen.x 104, 268–294. Curtiss, S., 1977. Genie: A Psycholinguistic Study of a Modern-day ‘Wild Child’. Academic Press, New York. Deese, J., 1969. The association structure of some common English adjectives. In: Snider, J.M. (Ed.), Semantic Differential. Aldine Publishing Company, Chicago, pp. 218–228. Elbert, T., Pantev, C., Wienbruch, C., Rockstroh, B., Taub, E., 1995. Increased cortical representation of the fingers of the left hand in string players. Science 270, 305–307. Flor, H., Elbert, T., Knecht, S., et al., 1995. Phantom-limb pain as a perceptual correlate of cortical reorganization following arm amputation. Nature 375, 482–484.

Green, D., Swets, J., 1966. Signal Detection Theory and Psychophysics. Wiley, New York. Hamann, S.B., 1996. Implicit memory in the tactile modality: evidence from Braille stem completion in the blind. Psychol. Sci. 7, 284–288. Hollins, M., 1989. Understanding Blindness. Lawrence Erlbaum Associates, Publisher, Hillsdale. Hubel, D.H., Wiesel, T.N., 1977. Functional architecture of macaque monkey visual cortex. Proc. R. Soc. Lond. B 198, 1–59. Hull, T., Mason, H., 1995. Performance of blind children on digit-span tests. J. Visual Impair. Blin. 89, 166–169. Jonides, J., Kahn, R., Rozin, P., 1975. Imagery instruction improve memory in the blind subjects. Bull. Psychonomic Soc. 5, 424–426. Kaas, J.H., 2000. The reorganization of sensory and motor maps after injury in adult mammals. In: Gazzaniga, M.S. (Ed.), The New Cognitive Neurosciences. MIT Press, Cambridge, MA, pp. 223–236. Kirk, R.E., 1982. Experimental Design: Procedures for the Behavioral Sciences. BrooksyCole, Belmont, CA. Kool, V.K., Rana, M., 1980. Tactual short term memory of blind and sighted children. Psychologia 23, 173–178. Kujala, T., Alho, K., Kekoni, J., et al., 1995. Auditory and somatosensory event-related brain potentials in early blind humans. Exp. Brain Res. 104, 519–526. ¨¨ ¨ Kujala, T., Lehtokoski, A., Alho, K., Kekoni, J., Naatanen, R., 1997. Faster reaction times in the blind than sighted during bimodal divided attention. Acta Psychol. 96, 75–82. ¨¨ ¨ Kujala, T., Alho, K., Naatanen, R., 2000. Cross-modal reorganization of human cortical functions. Trends Neurosci. 23, 115–120. Lessard, N., Paree, M., Lepore, F., Lassonde, M., 1998. Earlyblind human subjects localize sound sources better than sighted subjects. Nature 395, 278–280. Marchant, B., Malloy, T., 1984. Auditory, tactile and visual imagery in PA learning by congenitally blind, deaf, and normal adults. J. Ment. Imagery 8, 19–32. Millar, S., 1982. Studies of the deaf and the blind. In: Burton, A. (Ed.), The Pathology and Psychology of Cognition. Methuen, London, pp. 135–168. Miller, L., 1992. Diderot reconsidered: visual impairment and auditory compensation. J. Visual Impair. Blin. 86, 206–210. Mills, A.E., 1988. Visual handicap. In: Bishop, D., Mogford, K. (Eds.), Language Development in Exceptional Circumstances. Churchill Livingstone, Edinburgh, pp. 150–163. Muchnik, C., Efrati, M., Nemeth, E., Malin, M., Hildesheimer, M., 1991. Central auditory skills in blind and sighted subjects. Scand. Audiol. 20, 19–23. Neville, H.J., Coffey, S.A., Lawson, D.S., Fischer, A., Emmorey, K., Bellugi, U., 1997. Neural systems mediating American sign language: effects of sensory experience and age of acquisition. Brain Lang. 57, 285–308. Paivio, A., Okovita, H.W., 1971. Word imagery modalities and associative learning in blind and sighted subjects. J. Verb. Learn. Verb. Behav. 10, 506–510.

¨ ¨ B. Roder, F. Rosler / International Journal of Psychophysiology 50 (2003) 27–39 Pantev, C., Oostenveld, R., Engelien, A., Ross, B., Roberts, L.E., Hike, M., 1998. Increased auditory cortical representation in musicians. Nature 393, 811–814. Pring, L., 1988. The ‘reverse-generation’ effect: a comparison of memory performance between blind and sighted children. Br. J. Psych. 79, 387–400. Pring, L., Freistone, S.E., Katan, S.A., 1990. Recalling pictures and words: reversing the generation effect. Curr. Psychol.: Res. Rev. 9, 35–45. Recanzone, G., 2000. Cerebral cortical plasticity: perception and skill acquisition. In: Gazzaniga, M.S. (Ed.), The New Cognitive Neurosciences. MIT Press, Cambridge, MA, pp. 237–247. ¨ Roder, B., Neville, H., 2003. Developmental functional plasticity. In: Grafman, J., Robertson, I. (Eds.), Handbook of Neuropsychology, vol. 9. Elsevier. ¨ ¨ ¨ Roder, B., Rosler, F., Hennighausen, E., Nacker, F., 1996. Event-related potentials during auditory and somatosensory discrimination in sighted and blind human subjects. Cogn. Brain Res. 4, 77–93. ¨ ¨ Roder, B., Rosler, F., Neville, H.J., 1999. Effects of interstimulus interval on auditory event-related potentials in congenitally blind and normally sighted humans. Neurosci. Lett. 264, 53–56. ¨ ¨ ¨ ¨ Roder, B., Teder-Salejarvi, W., Sterr, A., Rosler, F., Hillyard, S.A., Neville, H.J., 1999. Improved auditory spatial tuning in blind humans. Nature 400, 162–166. ¨ ¨ Roder, B., Rosler, F., Neville, H.J., 2000. Event-related potentials during language processing in congenitally blind and sighted people. Neuropsychologia 38, 1482–1502. ¨ ¨ Roder, B., Rosler, F., Neville, H.J., 2001. Auditory memory in congenitally blind adults: a behavioral-electrophysiological investigation. Cogn. Brain Res. 11, 289–303.

39

¨ ¨ Roder, B., Demuth, L., Streb, J., Rosler, F., 2002. Semantic and syntactic priming in auditory word recognition in congenitally blind adults. Language and Cognitive Processes, in press. ¨ ¨ Roder, B., Stock, O., Bien, S., Neville, H., Rosler, F., 2003. Speech processing activates visual cortex in congenitally blind adults. Eur. J. Neurosci. 16, 930–936. Roediger, H.L.I.I.I., McDermott, K.B., 1995. Creating false memories: remembering word not presented in lists. J. Exp. Psychol. wLearn. Mem. Cogn.x 21, 803–814. ¨ Ronnberg, J., Nilsson, L., 1987. The modality effect, sensory handicap and compensatory functions. Acta Psychol. 65, 263–283. Schacter, D.L., 1997. False recognition and the brain. Curr. Dir. Psychol. Sci. 6, 65–70. Schacter, D.L., Norman, K., Koutstaal, W., 1998. The cognitive neuroscience of constructive memory. Annu. Rev. Psychol. 49, 289–318. Thinus-Blanc, C., Gaunet, F., 1997. Representation of space in blind persons: vision as a spatial sense? Psychol. Bull. 121, 20–42. Wagner, A.D., Schacter, D.L., Rotte, M., et al., 1998. Building memories: remembering and forgetting of verbal experiences as predicted by brain activity. Science 281, 1188–1191. Warren, D.H., 1994. Blindness and Children: An Individual Differences Approach. Cambridge University Press, New York. Weber-Fox, C.M., Neville, H.J., 1996. Maturational constraints on functional specialization for language processing: ERP and behavioral evidence in bilingual speakers. J. Cogn. Neurosci. 8, 231–256. Weyerts, H., Tendolkar, I., Smid, H.G.O.M., Heinze, H.J., 1997. ERPs to encoding and recognition in two different inter-item association tasks. NeuroReport 8, 1583–1588.