تغییرات توجه در نابینای اولیه: مطالعه ERP فرایندهای کنترل توجه در صورت عدم وجود اطلاعات فضایی بصری

Abstract To investigate the role of visual spatial information in the control of spatial attention, event-related brain potentials (ERPs) were recorded during a tactile attention task for a group of totally blind participants who were either congenitally blind or had lost vision during infancy, and for an age-matched, sighted control group who performed the task in the dark. Participants had to shift attention to the left or right hand (as indicated by an auditory cue presented at the start of each trial) in order to detect infrequent tactile targets delivered to this hand. Effects of tactile attention on the processing of tactile events, as reflected by attentional modulations of somatosensory ERPs to tactile stimuli, were very similar for early blind and sighted participants, suggesting that the capacity to selectively process tactile information from one hand versus the other does not differ systematically between the blind and the sighted. ERPs measured during the cue–target interval revealed an anterior directing attention negativity (ADAN) that was present for the early blind group as well as for the sighted control group. In contrast, the subsequent posterior late direction attention negativity (LDAP) was absent in both groups. These results suggest that these two components reflect functionally distinct attentional control mechanisms which differ in their dependence on the availability of visually coded representations of external space.

1. Introduction Covert shifts of spatial attention can affect the perceptual processing of stimuli that are located within the current focus of attention. Evidence for such attentional modulations of sensory processing has been provided by event-related brain potential (ERP) studies, which have demonstrated that amplitudes of early modality-specific ERP components are enhanced when visual (e.g., Mangun & Hillyard, 1991), auditory (e.g., Näätänen, 1982) or tactile (e.g., Michie, Bearpark, Crawford, & Glue, 1987) stimuli are presented at attended relative to unattended locations. In contrast, the covert attentional control processes that are elicited in anticipation of task-relevant stimuli at specific locations, and which are responsible for spatially specific effects of attention, have only recently become the object of experimental investigation. Such control processes can be studied by using fMRI or ERP measures of brain activity in response to attentional cues that direct attention to one side versus the other, prior to the presentation of any sensory events at attended versus unattended locations (see Corbetta & Shulman, 2002, for a review of fMRI evidence for attentional control circuits). While most ERP studies have examined attentional control processes triggered during covert anticipatory shifts of visual spatial attention (e.g., Harter, Miller, Price, LaLonde, & Keyes, 1989; Yamaguchi, Tsuchiya, & Kobayashi, 1994; Hopf & Mangun, 2000; Nobre, Sebestyen, & Miniussi, 2000), some recent studies have now begun to look for ERP correlates of covert attentional orienting towards the anticipated side of relevant auditory or tactile events (Eimer et al., 2002, Eimer et al., 2003a, Eimer et al., 2003b and Eimer et al., 2004; Eimer & Van Velzen, 2002). These experiments have uncovered two lateralised ERP components that are elicited during the cue–target interval and which are sensitive to the direction of a cued attentional shift. An enhanced negativity at frontal electrodes contralateral to the side of attentional shifts between 300 and 500 ms after cue onset (‘anterior directing attention negativity’, ADAN) was followed by an enhanced contralateral positivity at posterior electrodes (‘late directing attention positivity’, LDAP), which emerged during later phases of the cue–target interval. Interestingly, these effects were not just triggered during shifts of visual attention, but also when attention was directed towards anticipated task-relevant auditory or tactile events (e.g., Eimer et al., 2002).1 The functional interpretation of such lateralised ERP components that are elicited during covert attention shifts is currently under debate. Based on the observation that these components are very similar during attentional shifts towards task-relevant visual, auditory or tactile events, we have previously suggested (Eimer & Driver, 2001; Eimer et al., 2002) that they reflect the activity of multimodal attentional control processes, which determine the spatial parameters of attentional shifts in a supramodal fashion, regardless of sensory modality (see also Farah, Wong, Monheit, & Morrow, 1989, for similar arguments). While the ADAN may reflect processes within a multimodal ‘anterior attention system’ (Posner & Petersen, 1990), the posterior LDAP might be linked to the activation of posterior parietal areas, which are known to be involved both in the orienting of spatial attention (e.g., LaBerge, 1995) and in the integration of information from different sense modalities (e.g., Andersen, Snyder, Bradley, & Xing, 1997). In contrast, others (e.g., Harter et al., 1989) have argued that the posterior LDAP component reflects the spatially selective activation of modality-specific visual areas in anticipation of task-relevant visual events. This hypothesis might appear inconsistent with recent observations that this component is elicited not only when attention is allocated to the expected location of visual stimuli, but also during shifts of tactile or auditory attention (Eimer et al., 2002 and Eimer et al., 2003a; Eimer & Van Velzen, 2002; see also Green, Teder-Sälerjärvi, & McDonald, 2005), since there is no obvious reason to assume that visual areas should be selectively activated in anticipation of auditory or tactile events. However, if one assumes that the control of attentional shifts is generally dominated by visual spatial information, even when other modalities are task-relevant, the possibility that ADAN and/or LDAP might predominantly reflect visual attentional control becomes more plausible. Vision provides superior spatial acuity relative to hearing or touch, thus allowing for more precise tuning of spatial attention. In view of this, it could also be used to control shifts of attention to anticipated locations of auditory or tactile events. If visual information was used to guide the spatial selection of auditory or tactile stimuli, lateralised ERP components elicited during attentional shifts towards anticipated tactile or auditory events might primarily reflect shifts of attention within visual space, rather than the activity of a genuinely multimodal attentional control system. On a more general level, the question under debate is which spatial reference frames are used when shifts of attention are programmed and executed. Attentional orienting might be based, primarily or exclusively, on visually mediated representations of external space, even when modalities other than vision are currently task-relevant. Alternatively, the control of spatial attention might be based on multiple frames of reference, including coordinates of visually represented external space, body-centred space, somatotopic space, or, in the case of genuinely supramodal attentional control, amodal spatial coordinates. Thus, it is clearly important to investigate whether the lateralised ERP components that are triggered during cued shifts of spatial attention (ADAN and LDAP) reflect attentional control processes that are based on a single shared spatial frame of reference, or whether these components are linked to separable control mechanisms that differ in terms of their spatial coordinates. Some initial evidence for the latter hypothesis comes from previous ERP studies demonstrating dissociations between the ADAN and LDAP. In one experiment (Eimer et al., 2003a) participants directed attention to their left versus right hand (as indicated by a central precue on every trial), and ERPs were recorded during the cue–target interval under conditions where hands were either uncrossed or crossed. This manipulation of hand posture had a marked effect on the ADAN component. When considered in terms of the direction of attentional shifts in external space, the ADAN was delayed and reversed polarity with crossed relative to uncrossed hands, suggesting that the attentional control processes reflected by the ADAN may be primarily based on somatotopic spatial coordinates, and not on visually defined external space. In contrast, when considered in terms of external space, the LDAP component was completely unaffected by crossing the hands, thus indicating that the control processes reflected by this component operate primarily on the basis of representations of visually mediated external space. Another study (Eimer et al., 2004) supported these conclusions by demonstrating that when participants were cued to direct attention to the left or right hand for a tactile task, the distance between hands in external space modulated the LDAP (which was more pronounced when hands were wide apart), but left the ADAN component entirely unaffected. One way to investigate the hypothesis that the LDAP (but not the ADAN) reflects the visually mediated control of attention shifts is to eliminate continuously available ambient visible sources of information about task-relevant stimulus locations. When visual spatial information about the visible positions of hands and arms, or the visible location of tactile and auditory stimulators, is eliminated either by blindfolding participants or by running an experiment in the dark, and visual cues are no longer available to aid and possibly dominate the spatial selection of tactile or auditory events, lateralised ERP components linked to attentional control based on coordinates of visual space should be absent, whereas components that are based on other non-visual spatial reference frames should remain unaffected. We first tested this prediction in a study (Eimer et al., 2003b) where participants had to shift their attention towards their left or right hand (as indicated by an auditory cue presented at the start of each trial) in anticipation of task-relevant tactile events, either in a normally lit environment or in darkness. The ADAN component, which was present under normal illumination conditions, remained virtually unchanged in darkness. Although the LDAP was also reliably present in both conditions, its amplitude was reduced in the dark. The finding that the ADAN does not seem to depend on the continuous availability of visual spatial information is in line with the idea that this component reflects the control of attention within somatotopic or body-centred space. However, the observation that the LDAP was reduced, but still reliably present in the dark, does not allow any firm conclusions with respect to its dependence on visual spatial representations. One account of the presence of an attenuated LDAP during shifts of attention in the dark assumes that the brain might still utilize stored visual representations of external space, or visual imagery, in the control of tactile attention shifts in the dark. In our previous experiment (Eimer et al., 2003b), all participants had received several training blocks under normal illumination conditions, with half of them being tested first in a lit experimental room before being tested in the dark. Thus, they may have been able to use stored visual representations of the spatial layout of the task situation when directing tactile attention in the dark. Consistent with this explanation, LDAP amplitudes were found to be even smaller in a follow-up experiment where all blocks were conducted in darkness (although training blocks were still run in a lit environment; see Eimer et al., 2003b, for more details). The main aim of the present experiment was to further clarify the role of visual spatial information during attentional orienting and its impact on lateralised ERP components triggered during shifts of spatial attention. ERP correlates of tactile attention shifts were measured in 10 early blind participants who were either blind from birth or had lost all vision within the first 2 years of life (see Table 1 for details). Due to the lack of visual input during perceptual and cognitive development, congenitally or early blind participants clearly cannot employ a visually defined spatial frame of reference, or visual imagery, to guide shifts of attention towards anticipated task-relevant tactile events. Investigating whether lateralised ERP components (ADAN and LDAP) are present during cued shifts of tactile attention in early blind participants2 therefore provides a strong test as to whether one or both of these components reflect attentional control processes that are primarily guided by visual information. Control processes that depend on visually defined spatial coordinates should be entirely absent during attentional orienting in the early blind. In contrast, control processes (and their ERP correlates) which operate independently of visual information should remain present. Table 1. Description of early blind participants No. Age Gender Handedness Visual perception Age of onset Cause of blindness 1 60 Male Right None 18 months Eye glass damage 2 54 Female Right None Birth Retinopathy of prematurity 3 63 Male Right Diffuse light Birth Nervus opticus atrophy 4 54 Male Right None Birth Retinopathy of prematurity 5 23 Female Neither Diffuse light Birth Retinopathy of prematurity 6 23 Female Neither Diffuse light Birth Retinopathy of prematurity 7 29 Female Right None 2 years Detached retina 8 54 Female Right None Birth Retinopathy of prematurity 9 55 Male Neither None Birth Retinopathy of prematurity 10 50 Male Right None 13 months Detached retina Table options In addition to measuring ERPs during tactile attention shifts for early blind participants, we also tested a group of 10 age-matched sighted control participants. Both sighted and blind participants had to detect and respond to infrequent tactile target stimuli when these were delivered to the currently task-relevant hand, whilst ignoring tactile stimuli when these were presented to the other irrelevant hand. Experimental blocks were conducted in a completely dark experimental booth. Similar to our previous study (Eimer et al., 2003b), a trial-by-trial cueing paradigm was employed where the relevant hand was indicated at the beginning of each trial by an auditory cue. In contrast to this previous study, sighted control participants received task instructions and all training blocks needed to familiarize them with task procedures in the dark. This was to eliminate the possibility that, during training, they would build up visual representations of the spatial layout of the experimental set-up, which might be utilized to guide shifts of tactile attention in the dark. One set of analyses was conducted for ERP components elicited in the cue–target interval in response to auditory cues directing tactile attention to the left versus right hand. Based on our earlier finding with sighted participants, which had shown that the ADAN component appears to be entirely unaffected by the absence of ambient visual information (Eimer et al., 2003b), we expected this component to be present for the sighted control group. The new question was whether an ADAN would also be elicited in the early blind group. If this component was linked to attentional control processes that depend at least in part upon the availability of visual input during perceptual and cognitive development, it might be attenuated or even absent in the early blind. The other question concerned the fate of the posterior LDAP. If this component reflected the activity of attentional control mechanisms that rely on the present or past availability of visual spatial input, this component should be entirely absent during attentional orienting in early blind participants. For the sighted control group, the question was whether an LDAP would still be elicited during shifts of tactile attention even though participants had no opportunity to build up and store visual spatial representations of the task situation. Another set of analyses was conducted for somatosensory ERP components triggered in early blind and sighted participants in response to tactile non-target stimuli presented to the cued (attended) or uncued (unattended) hand. With sighted participants, directing tactile attention to one hand versus the other has been found to result in an enhancement of the somatosensory N140 component, which is usually followed by a sustained attentional negativity beyond 200 ms post-stimulus (Michie et al., 1987 and Eimer and Forster, 2003). There is substantial evidence that the early loss of visual information can result in compensatory improvements of spatial perception in the remaining intact modalities (see Röder & Neville, 2003, for a review). For example, Röder et al. (1999b) have demonstrated superior auditory localization abilities for congenitally blind adults when attending to sounds in peripheral auditory space. Such compensatory changes might in principle be reflected by earlier, or more pronounced effects of spatial attention on somatosensory ERPs for the early blind as compared to the sighted group. However, the few previous ERP studies to date that have investigated this issue found little evidence for an improvement of spatially selective attentional processing in the early blind, relative to sighted people. When comparing sighted and blind participants in terms of the effects of sustained spatial attention on auditory ERPs (Liotti, Ryder, & Woldorff, 1998), or on somatosensory as well as auditory ERPs (Hötting, Rösler, & Röder, 2004), no indication of earlier or more pronounced attentional modulations of early modality-specific ERP components were observed for the blind. The present experiment investigated this issue by contrasting the effects of tactile-spatial attention on somatosensory ERPs for early blind and sighted participants under transient attention conditions where attentional orienting was cued on a trial-by-trial basis.

3. Results 3.1. Behavioural performance Mean vocal response times to cued tactile targets were 545 ms in the blind group, and 595 ms in the sighted group. This difference failed to reach significance (t(18) < 1). In the blind group, responses to tactile targets presented to the right hand were faster than responses to left hand targets (533 ms versus 557 ms), and this difference was significant (t(9) = 2.3; p < .05). No such difference was present for the sighted controls (595 and 594 ms for left and right targets). False alarms to tactile non-target stimuli were present on 0.6% (blind group) and 0.8% (sighted group) of all non-target trials, and this difference was not significant. Blind participants missed 6.3% of all targets presented at cued locations, and produced false alarms on 6.9% of all trials where tactile targets were presented at uncued locations, as compared to 9.2% misses and 2.1% false alarms for the sighted group. Based on these data, sensitivity measures (d′) and measures of response bias (C) were computed for both groups (derived from signal detection theory, and described in Macmillan & Creelman, 1991). Whereas d′ did not differ significantly between the two groups, C was significantly larger for sighted as compared to blind participants (t(18) = 4.3; p < .001), demonstrating that the blind used a more liberal response criterion. 3.2. ERP correlates of tactile-spatial orienting in the interval between auditory cues and peripheral tactile stimuli Fig. 1 and Fig. 2 show ERPs elicited at lateral electrodes over the left and right hemisphere in the interval between cue onset and onset of the subsequent tactile stimulus, displayed separately for auditory cues directing tactile attention to the left side (solid lines) and to the right side (dashed lines). As can be seen from these figures, the pattern of ERP lateralisations sensitive to the direction of a cued attentional shift was remarkably similar across both groups. For blind as well as sighted participants, a negativity contralateral to the direction of an attentional shift (anterior directing attention negativity) was maximal at frontocentral electrodes, but also seemed to be present at more posterior sites (CP5/6). Importantly, no contralateral late directing attention positivity appeared to be present for either group. Grand-averaged ERPs elicited for congenitally blind participants over the left ... Fig. 1. Grand-averaged ERPs elicited for congenitally blind participants over the left and right hemisphere in the interval between cue onset and onset of the subsequent peripheral tactile stimulus. ERPs in response to auditory cues directing attention to the left side (solid lines), and cues directing attention to the right side (dashed lines) are shown separately. An anterior directing attention negativity (ADAN) was elicited at frontal and centroparietal sites. Figure options Grand-averaged ERPs elicited for the sighted control group over the left and ... Fig. 2. Grand-averaged ERPs elicited for the sighted control group over the left and right hemisphere in the interval between cue onset and onset of the subsequent peripheral tactile stimulus, in response to auditory cues directing attention to the left side (solid lines), and cues directing attention to the right side (dashed lines). As for the blind group, an anterior directing attention negativity (ADAN) was present. Figure options The presence and time course of the ADAN component, its similarity across blind and sighted participants, and the absence of a posterior LDAP component in both groups is further illustrated in Fig. 3, which shows difference waveforms obtained at lateral anterior (top panels), central (middle panels) and posterior electrode pairs (bottom panels), for the blind group (solid lines) and the sighted group (dashed lines). These difference waves were obtained by first subtracting ERPs recorded during attentional shifts to the right from ERPs elicited during leftward attentional shifts, and then subtracting the resulting difference waveforms at right electrodes from the difference waveforms emerging at corresponding electrodes over the left hemisphere. In the resulting double subtraction waveforms, an overall negativity contralateral to the direction of attentional shifts (ADAN) is reflected by positive amplitude values (downward-going deflections). Any contralateral positivity (LDAP) would have been reflected by negative values (upward deflections). Fig. 3 shows that an ADAN component was elicited at about 300 ms following cue onset in both groups. With the possible exception of F7/8, where the ADAN seems more pronounced in the sighted control group, overall this component appears to be similar in amplitude for both groups and remained present at lateral anterior and central sites throughout the cue–target interval. In contrast, there was no evidence that a posterior LDAP component was elicited during later stages of the cue–target interval in either the blind or sighted group. Difference waveforms obtained at lateral anterior (top), central (middle) and ... Fig. 3. Difference waveforms obtained at lateral anterior (top), central (middle) and posterior (bottom) electrodes in the 700 ms interval between cue onset and onset of the subsequent peripheral tactile stimulus, illustrating the time course of lateralised ERP modulations sensitive to the direction of attentional shifts for blind participants (solid lines) and for the sighted control group (dashed lines). Difference waveforms were generated by first subtracting ERPs in response to cues directing attention to the right from ERPs in response to cues directing attention to the left; and then subtracting the resulting difference waves at right electrodes from the difference waveform obtained for the corresponding left-hemisphere electrode. Enlarged negativities contralateral to the direction of attentional shifts are reflected by positive amplitude values (downward-going deflections). Enhanced contralateral positivities would have been reflected by negative values (upward-going deflections). Waveforms show a sustained anterior directing attention negativity (ADAN) for both groups, but no evidence for any posterior late directing attention positivity (LDAP) in either group. Figure options Statistical analyses confirmed these informal observations. ERPs elicited during leftward and rightward attention shifts were compared directly as a function of the recording hemisphere, prior to the double subtraction visualised in Fig. 3. No systematic ERP modulations sensitive to the direction of attentional shifts were observed within the first 300 ms following cue onset. In the 300–500 ms post-cue interval, significant hemisphere × cued direction interactions were present at anterior sites (F(1,18) = 20.4; p < .001), as well as at central sites (F(1,18) = 18.1; p < .001), reflecting the enhanced negativity contralateral to the direction of an attentional shift (ADAN) visible in Fig. 1, Fig. 2 and Fig. 3. Importantly, there were no indications of any group × hemisphere × cued direction interactions at lateral anterior or central sites (both F(1,18) < 1), indicating that there were no systematic differences between ADAN components triggered during shifts of tactile attention in the blind and sighted groups. Analyses conducted separately for each group confirmed the presence of significant hemisphere × cued direction interactions at lateral anterior and lateral central recording electrodes in both groups (all F(1,9) > 8.1; all p < .02). No overall significant hemisphere × cued direction interaction, or any group × hemisphere × cued direction interaction was present in the 300–500 ms interval at lateral posterior electrodes. A similar pattern of results was found in the subsequent 500–700 ms post-cue interval (corresponding to the final 200 ms prior to the onset of a lateral tactile stimulus). Again, significant hemisphere × cued direction interactions were present at lateral anterior (F(1,18) = 42.7; p < .001) as well as at lateral central electrodes (F(1,18) = 28.8; p < .001), demonstrating that the ADAN remained present during the later phase of the cue–target interval. As was the case for 300–500 ms time window, no sign of any group × hemisphere × cued direction interaction was found at these electrode sites (both F(1,18) < 1.8). Again this strongly suggests that ADAN amplitudes did not differ systematically across the blind and sighted groups. 3 Analyses conducted separately for each group confirmed that significant hemisphere × cued direction interactions were present at lateral anterior sites as well as at lateral central sites in both groups (all F(1,9) > 11.5; all p < .01). Importantly, in contrast to previous investigations of cued shifts of spatial attention with sighted participants, there was no statistical evidence whatsoever for the presence of a posterior LDAP component during this 500–700 ms measurement interval. At lateral posterior electrodes, the overall hemisphere × cued direction interaction failed to reach significance (F(1,18) < 1.5). A hemisphere × cued direction × electrode site interaction (F(2,36) = 7.1; p < .02; ɛ = .837) was accompanied by a significant hemisphere × cued direction interaction at P3/4 (F(1,18) = 9.4; p < .01), reflecting the fact that the ADAN continued to be present, albeit in attenuated fashion, at this electrode pair (see Fig. 3). Importantly, however, there was no trace of any hemisphere × cued direction interaction at lateral posterior electrode pairs P3/4 and OL/R (both F(1,18) < 1), where the LDAP component was reliably found in previous studies of spatial orienting in sighted participants. In addition, not only were group × hemisphere × cued direction interactions not found at either of these electrode pairs (both F(1,18) < 1), but also follow-up analyses conducted separately for the blind and sighted groups failed to find any indication of hemisphere × cued direction interactions at P3/4 and OL/R, thereby strongly suggesting that the LDAP component was not only absent in the early blind group, but also for the sighted controls. 3.3. Effects of spatial attention on somatosensory ERPs ERPs elicited in response to tactile non-target stimuli at cued locations (solid lines) and uncued locations (dashed lines) are shown in Fig. 4 for blind participants, and in Fig. 5 for the sighted control group. Waveforms are displayed separately for midline electrodes and for electrode sites contralateral (left panels) and ipsilateral (right panels) to the side of tactile stimulation. In both groups, somatosensory N140 components appear to be enhanced when tactile stimuli were presented to the attended hand relative to tactile stimuli presented to the unattended hand. In addition, a sustained enhanced negativity in response to attended relative to unattended tactile stimuli was elicited in a similar fashion for sighted and early blind participants. Grand-averaged somatosensory ERPs elicited for congenitally blind participants ... Fig. 4. Grand-averaged somatosensory ERPs elicited for congenitally blind participants at midline electrodes, and at sites contralateral (C) and ipsilateral (I) to the side of stimulus presentation, in response to tactile non-target stimuli at cued locations (solid lines) and uncued locations (dashed lines) in the 500 ms interval following stimulus onset. Figure options Grand-averaged somatosensory ERPs elicited for the sighted control group at ... Fig. 5. Grand-averaged somatosensory ERPs elicited for the sighted control group at midline electrodes, and at sites contralateral (C) and ipsilateral (I) to the side of stimulus presentation, in response to tactile non-target stimuli at cued locations (solid lines) and uncued locations (dashed lines) in the 500 ms interval following stimulus onset. Figure options No significant main effects of attention or group × attention interactions were present for the P100 component (90–120 ms post-stimulus). In the N140 latency range (130–170 ms post-stimulus), main effects of attention were obtained at contralateral and ipsilateral electrodes as well as at midline sites (all F(1,18) > 13.8; all p < .05), demonstrating that directing tactile attention to one hand versus the other modulated N140 amplitudes. A main effect of group was significant at ipsilateral electrodes (F(1,18) = 6.5; p < .02), but failed to reach significance at contralateral and midline sites. This reflects the fact that somatosensory ERPs in the N140 time range tended to be generally more positive for the blind relative to the sighted group (see Fig. 4 and Fig. 5). However, and more importantly, no group × attention interactions were present in the N140 time window (all F(1,18) < 1), suggesting that analogous attentional N140 modulations were elicited in the blind and sighted groups. Follow-up analyses conducted separately for both groups revealed reliable effects of attention on N140 amplitudes at ipsilateral, contralateral, and midline sites for the sighted group (all F(1,9) = 5.1; all p < .001), and reliable attentional effects at ipsilateral sites (F(1,9) = 11.1; p < .01) and at midline electrodes (F(1,9) = 5.7; p < .05) for the blind group. In the 200–350 ms measurement window, main effects of attention were obtained at contralateral, ipsilateral, as well as midline sites (all F(1,18) > 39.5; all p < .001), reflecting the sustained enhanced negativity for attended relative to unattended tactile stimuli shown in Fig. 4 and Fig. 5. Main effects of group were also present at these sites (all F(1,18) > 7.1; all p < .02), as ERPs were generally more positive in the blind as compared to the sighted group during this time window. However, analogous to the results found for the N140 component, there was no indication of any group × attention interaction (all F(1,18) < 1) suggesting that attentional ERP modulations triggered between 200 and 350 ms post-stimulus were comparable in size across blind and sighted participants. Fig. 4 and Fig. 5 suggest that at longer latencies, this attentional negativity might extend to more posterior (occipital) sites, particularly in the early blind group. To investigate this, additional post hoc analyses were conducted for ERP waveforms obtained at Oz, OL and OR between 300 and 400 ms after stimulus onset. A main effect of attention was obtained for the blind group (F(1,9) = 9.0; p < .02), although this effect was not significant for sighted participants (F(1,9) = 2.5; p < .15). However, this difference was not substantiated by an overall significant group × attention interaction.