Local-global interference is modulated by age, sex and anterior corpus callosum size

1.1. Lateralized processing and age

The competition hypothesis is further supported by neuroimaging studies reporting smaller corpora callosa with older age (Pfefferbaum et al., 2000; Salat et al., 2005; Sullivan et al., 2006) and bihemispheric activation patterns even in less complex tasks that were unilaterally processed in young participants (Cabeza et al., 2002; Reuter-Lorenz et al., 2000). In a meta-analysis of three separate fMRI studies, Konishi and colleagues (2001) found that additional activity in the less-relevant hemisphere occurred early during processing but then disappeared, which is indicative of an active inhibition by the task-relevant/dominant hemisphere. In older subjects, bilateral involvement may arise when callosal inhibitory interactions responsible for maintaining hemispheric dominance of particular brain functions are degraded (Kinsbourne, 1974; Kinsbourne and Byrd, 1985). Examples include local-global interference, distractor incongruency and attentional selection of task-relevant from irrelevant information, all of which have also been associated with frontal cortex functions (Banich et al., 2000; MacDonald et al., 2000; Weissman et al., 2003, 2005). Given that aging is associated with selective degradation of anterior callosal subregions as assessed by means of Magnetic Resonance Imaging (Raz et al.,1995; Salat et al., 2001; Sullivan et al., 2002) and MR diffusion tensor imaging (Pfefferbaum and Sullivan, 2003; 2005; Pfefferbaum et al., 2000, 2005; Salat et al., 2005; Sullivan et al., 2001), processes of the executive attention system, i.e., local-global interference, may be particularly affected in older participants (Anderson et al., 2000; Cabeza, 2001, 2002; Cabeza et al., 1997).

Alternatively to the explanation that recruitment of additional brain areas (in the less-relevant hemisphere) results from less callosal inhibition in older participants (competition hypothesis) is the possibility that additional recruitment serves a compensatory function to overcome age-related cognitive decline (cooperation hypothesis) (Cabeza et al., 1997; Madden et al., 1999; McIntosh et al., 1999; Reuter-Lorenz et al., 2000). Compensatory recruitment of additional brain areas likely requires integrity of the corpus callosum, and smaller corpora callosa would be associated with poorer performance. The cooperation hypothesis is supported by reaction time studies showing that older participants profited more from bilateral stimulation conditions, requiring callosal transfer, than from unilateral conditions (Reuter-Lorenz and Stanczak, 2000; Reuter-Lorenz et al., 1999), and by functional imaging studies showing bilateral activation in high but not low performing older adults (Cabeza et al., 2002; Rosen et al., 2002). In addition, whether callosal interhemispheric interaction is inhibitory or cooperative may likely depend on task demands, task complexity and the specific process (e.g., inhibition vs. facilitation) and lateralization of functions (e.g., verbal-spatial, local-global).

1.2. Interference and facilitation in global-local processing

Interference may emerge from at least three sources: the stimulus content at global and local levels, i.e., when local and global stimuli are incongruent compared with when they are congruent (Kotchoubey et al., 1997; Proverbio et al., 1998); from response conflict when feature processing at the unattended level suggests a different response from the attended stimulus attribute (Hübner and Malinowski, 2002; Malinowski et al., 2002; Volberg and Hübner, 2004; Weissman et al., 2003); or from sequential processing when the target location switches between local and global levels in two consecutive trials (level switching) (Han et al., 2000a; Schatz and Erlandson, 2003). Behaviorally, interference effects are usually defined as costs in response time (RT) and facilitation effects as benefits in RTs. Facilitation may emerge from (a) concurrent processing of matching local-global information, e.g., when the unattended level carries a target in addition to a target at the attended level (response facilitation), and (b) from sequential processing of target level repetition (repetition priming). In this procedure, the hierarchical level of the target (trial n) is considered in conjunction with the target level of the prior stimulus (trial n-1) (Kim et al., 1999; Kinsbourne and Byrd, 1985; Slavin et al., 2002). Such sequential priming effects, measured by the level-repetition procedure, are largely independent from effortful attention manipulation biasing perception to the global or local level (Lamb and Yund, 1996; Lamb et al., 1998, 2000). For example, Lamb and colleagues (2000) showed that pretrial cues that provided information about the level of the upcoming target affected performance with RT benefits for valid and costs for invalid cues relative to noninformative neutral cues. Facilitation effects from sequential priming were assumed to occur relatively early, at a postsensory but preattentive stage (Han et al., 2000a; Kim et al., 1999; Lamb et al., 1998, 2000; Robertson, 1996; Schatz and Erlandson, 2003), whereas interference effects often have been associated with later response-related stages (Han et al., 1999; Kaufmann et al., 2005; van Veen and Carter, 2005).

1.3. The role of the corpus callosum in interference and facilitation processes

A possible neural mechanism for facilitation and interference between global and local processing levels involves interhemispheric interactions (Corballis et al., 2004; Forster and Corballis, 2000). Evidence from a split-brain study (Robertson et al., 1993) suggests that the corpus callosum mediates local-global interference. Degradation of connectivity between brain areas occurs with age (postmortem evidence: Aboitiz et al., 1992; Kemper, 1994) and involves the corpus callosum and its cortical connections, especially in prefrontal areas (in vivo evidence: Madden et al., 2004; Pfefferbaum et al., 2005; Raz et al., 1997, 2000; Salat et al., 1997; Sullivan et al., 2006). Callosal thinning without disconnection, determined with MRI, can contribute to impairment in tasks requiring interhemispheric sensorimotor processing (Schulte et al., 2004). Further, slowing of interhemispheric transfer time in simple tasks in older (60 – 82 years) compared with younger participants has been related to age-related callosal thinning (Hopper et al., 1994; Janowsky et al., 1996; Jeeves and Moes, 1996; Thompson et al., 2000).

The purpose of this study was to test the hypothesis that age affects the corpus callosum and influences component processes susceptible to local-global interference and facilitation. If a larger callosum in younger age enables faster interhemispheric communication, then the two hemispheres may readily collaborate when solving complex tasks (Banich and Belger, 1990; Banich et al., 1990); similarly, larger callosal size would facilitate global/local level cooperation. Alternatively, inhibitory callosal projections may protect lateralized local/global processes, and greater callosal size would be associated with more robust inhibition of task-irrelevant information, thereby reducing interference. Thus, a larger CC would be related to better performance emerging from faster interhemispheric cooperation or from more robust inhibition of task-irrelevant information. Slower interhemispheric communication would be related to older age, and thinning of callosal size would be related to less facilitation and greater interference between global and local levels and greater difficulty in switching between levels (Lux et al., 2004; Miller et al., 2000; Robertson et al., 1993; Yazgan et al., 1995).

We devised a novel paradigm of global/local stimuli to probe the effects of age and regional callosal size on these different processing levels and attentional demands. Accordingly, this study tested 1) the contribution of regional callosal size to component hierarchical stimulus processes, 2) whether age or sex modulates callosal-hierarchical processing relations, and 3) whether anterior (genu), medial (body), and posterior (splenium) callosal pathways are differentially related to components of hierarchical stimulus processing defined by competition vs. cooperation between local and global levels (interference, response conflict and switching vs. response facilitation and repetition priming) and attentional demands (selective, divided).

2.1. Corpus callosum size, sex and age

Age correlated significantly with callosal genu area (r = −.43, p < .004, one-tailed), but not body (r = −.17, ns, one-tailed) or splenium (r = .01, ns) area (Figure 2). Although callsoal subareas did not significantly differ between women and men (MANOVA Pillai’s Trace F(3,33) = 1.26, ns; genu F(1,35) = 0.31, ns; body F(1,35) = 0.33, ns; splenium F(1,35) = 1.47, ns), women tended to have larger splenia (ICV-corrected Z-scores) than men relative to the genu and body areas (sex-by-CC area interaction F(1,35) = 3.42, p = .073; repeated measures ANOVA) (Figure 1).

2.2. Specific effects of local-global processing and relation to callosal area

3.1. Precedence effects

The precedence effect denotes which spatial scale – global or local – is processed faster and depends on task characteristics influencing stimulus salience or presentation time (Bruyer and Scailquin, 2000; Hübner and Malinowski, 2002). Shorter RTs to local than global targets indicated a local precedence effect. Factors that can potentially facilitate a local processing bias are stimulus characteristics (Boles and Karner, 1996; Ivry and Robertson, 1998; Lamb and Robertson, 1988, 1989; Wilkinson et al., 2001) and task demands (Kimchi, 1992). In contrast to other studies that used time constraints for stimulus presentation and reported global precedence effects (Navon, 1977; Slavin et al., 2002; Weissman et al., 2002), stimuli in the present study remained on the screen until the subject pressed the response button. This procedure ensured sufficient processing time for older participants, who commonly have prolonged RTs (Cabeza, 2001; Madden et al., 1996; Salthouse, 1996) as they did herein, but may also have afforded time to process beyond the global level. Further, in our paradigm black local letters were superimposed on a gray global letter presented on white background. This set of contrasts possibly led to changes in the spatial frequency spectrum (Ivry and Robertson, 1998) to more contrast balanced stimuli that produced larger effects on global than local processing and resulted in local precedence (Han et al., 2002; Lamb and Yund, 1993, 1996). Finally, Kimchi (1992) observed that global precedence becomes more likely with peripheral stimulus presentation and is less likely when stimuli are presented centrally, as they were in the present design.

The local precedence effect was significantly more pronounced with advancing age, consistent with studies that found local precedence effects in older participants (Bruyer and Scailquin, 2000; Polster and Rapcsak, 1994; Slavin et al., 2002). Moreover, a smaller genu area was associated with a greater local precedence effect. Age and genu area together (but not independently) predicted the local precedence effect. One possible explanation is that the corpus callosum insulates lateralized local-global processes and anterior callosal decline with older age and leads to a less robust inhibition permitting interference, particularly from the faster (local) process, to influence precedence effects. Alternatively, the genu may serve to integrate holistic stimuli rather than to insulate lateralized local-global processes. Local stimuli may be available to both hemispheres, whereas interhemispheric integration may be needed for the perception of global stimuli when requiring visual exploration. A decline in callosal function with older age would mean that holistic stimuli are not well integrated interhemispherically, thus, global function would decline more than local function. Finally, structural genu area and functional global/local processing are affected by a common factor, i.e., age, and it is possible that genu size is more an index of aging than of a processing bias leading to local over global precedence.

Correlations between age and genu area with the local precedence effect were evident in selective but not divided attention condition. However, the longer RTs in the divided compared with the selective attention conditions suggest greater task difficulty when participants have to direct their attention to the local and the global level than when they can focus on one level. Thus, the greater local precedence effect may be a result of greater task difficulty. One may further assume bilateral processing and callosal involvement with greater difficulty, especially in older participants, to recruit compensatory brain reserve and overcome age-related cognitive decline (Cabeza et al., 2002; Reuter-Lorenz et al., 1999, 2000) (compensatory hypothesis). We did not, however, find such a relationship. One possible explanation is inadequate statistical power. Another possibility is that global and local information are processed in parallel intrahemispherically and that the genu-precedence relationship under selective attention conditions reflects interhemispheric communication from top-down selective attentional control rather than from precedence processing itself. Thus, callosal inhibitory functions would serve top-down attentional control mechanisms when selectively directing attention to the relevant level and suppressing the irrelevant level; such gating functions would not be required when attention is directed to both levels simultaneously (Figure 1).

3.2. Interference from incongruent local-global information and response conflict

Our paradigm permitted testing of two types of interference: 1) interference from incongruent local-global information and 2) interference from response conflict created by stimuli for which information at each level was associated with a different response. As expected, both types of interference resulted in prolonged responses. Interference from local targets while attending the global level correlated with age and independently with genu size. The anterior cingulate cortex is involved in the regulation of attention and in monitoring response conflict (Botvinick et al., 2001, 2004; Carter et al., 1998), and is thought to detect conflicts occurring at later or response-related processing levels and signals the need to engage top-down attentional processes (van Veen and Carter, 2002; van Veen et al., 2001; Weissman et al., 2003). Our findings possibly add to identification of neural substrates of such attentional processes by demonstrating the influence of age-related genu thinning on interference control. The special role that this callosal region may play is in the preservation of lateralized local-global processes that become less robust with age. Callosal inhibition from age-related genu-thinning would allow interference from task-irrelevant information. Sex was an additional modifier of interference, in that ‘callosal structure - interference function’ relationships were evident in women but not in men. This observation adds to lateralization research demonstrating that functional cerebral asymmetry of visuospatial processes is more pronounced in men and tends to be more symmetrical in women (Corballis and Sidey, 1993; Hausmann and Güntürkün, 1999; McGlone, 1980; Rasmjou et al., 1999). Thus, in women, a smaller callosal size with older age may compromise mechanisms recruited to reduce interference using interhemispheric processes, whereas such interference may be resolved unilaterally in men.

Although one component of response conflict may be thought to be related to the integrity of motor areas and their callosal connections, we did not observe any correlations based on our measure of the callosal body. Further subdivision of the callosal body (Seltzer and Pandya, 1986) using marcostructural or microstructural neuroimaging methods of fiber integrity (Hofer and Frahm, 2006; Robertson et al., 1991) may be useful for future research on motor components of local-global processes.

No relations emerged among interference measures in the local instruction block, callosal area, and age. One possible explanation is that processing of the dominant level must be suppressed when processing at the nondominant level is required, and that the corpus callosum mediates such interhemispheric equilibration processes depending on task demands (Lux et al., 2004). Suppression of local precedence was also associated with smaller genu area, suggesting that with advancing age smaller anterior callosal volume increases local interference (global block) and amplifies an unbalanced inhibition pattern.

3.3. Response facilitation

Response facilitation effects, i.e., shorter RTs when targets appear at attended and unattended levels than when targets appear at attended levels only, were greater for local than global unattended targets and were independent of age and callosal size. Although the interpretation of non-significant results is problematic, one possibility is that facilitation effects from targets at both levels may originate intrahemipherically because each hemisphere is capable of processing local and global information but with different levels of efficacy (Robertson and Lamb, 1991; Yamaguchi et al., 2000).

3.4. Sequential repetition priming

Faciliation effects from repetition priming, i.e., shorter RTs to a trial (n) compared with the previous trial (n-1), were evident for both processing levels. Thus, identifying a target letter on a trial n-1 decreased identification time on the following trial n. This facilitation effect was higher for local than global targets, providing support for level-specific sequential priming (Kim et al., 1999). In the present study, repetition effects from global target repetition correlated with greater splenium area in men, whereas local target repetition effects correlated with smaller splenium area in women. The splenium connects posterior brain areas associated with primary visual and visuo-attentional processes. This finding supports the assumption of an early mechanism, perhaps at a postsensory but preattentive stage, in the visual processing stream (Kim et al., 1999; Lamb et al., 2000) that operates differently in men and women. One explanation for these sex effects in level-specific priming is that ‘left-hemispheric’ local processing advantages in women and ‘right-hemispheric’ global processing advantages in men are mediated by posterior callosal connectivity, here expressed as splenium size, which was proportionally larger in women than in men (Davatzikos and Resnick, 1998; DeLacoste-Utamsing and Holloway, 1982; Witelson, 1989).

In contrast to repetition priming, switch effects describe longer RTs to trial n than trial n-1 when the two trials contained targets at different levels. Cost effects from level switching emerged for both levels, although switching to the global level elicited higher costs than switching to the local level. According to the Attentional Print model (Robertson, 1996, 1999), information regarding the target level leaves a “print” that weights the high- and low-spatial-frequency channel to resolve whether targets are on a local or global level. The channel with the last resolved target receives a greater weight or “print” to resolve the following target, thus resulting in level-specific priming. With switch-level trials this weighting leads to a filtering of information through the “wrongly” primed channel that must be inhibited to resolve the target accurately. The local precedence effect in our study may reflect a higher weighting for the high-spatial-frequency channel that reduced the costs for switches towards the local level and enhanced costs for switches away from the local towards the global level. Lesion studies have associated the left inferior parietal cortex with level-switch effects (Robertson et al., 1988; Robertson and Rafal, 2000), but functional MRI studies in healthy participants report bilateral parietal (Wilkinson et al., 2001) and occipito-temporal involvement (Han et al., 2000a; Schatz and Erlandson, 2003). Although one would expect interhemispheric inhibition of wrongly primed spatial-frequency information of lateralized local-global processes via callosal structure, callosal size did not correlate with switch effects.

3.5. Conclusion

Interhemispheric functions underlying components of local-global processing are differentially affected by callosal size, age, and sex. According to the competition hypothesis, use of central stimulus presentations, which simultaneously provides each hemisphere with the same information, should lead to local-global competition such that each hemisphere assigns more resources to its specialized target level. In support of the competition hypothesis, precedence effects were related to callosal genu area under selective but not divided attention conditions. Resource assignment occurs when attention is directed to one target level but not when attentional resources are divided between local and global features. Anterior callosal size is also relevant in the presence of conflicting or incongruent local information while attention is directed to the global level. Averting such interference from local unattended targets requires the integrity of the anterior callosum, and both resistance to local interference and integrity of the anterior callosum decline with age. Response facilitation from repetition priming, however, correlates with splenium area regardless of age and differentially for men and women. Finally, our data indicate the following dissociation: Anterior callosal structure is associated with inhibitory processes (i.e., interference from local incongruent information, and response inhibition from conflicting local information while processing global information), which are vulnerable to the effects of age and sex, whereas posterior callosal structure is associated with facilitation processes from repetition priming dependent on sex but independent of age.

4.1. Participants

The study sample comprised 37 healthy adults (18 women and 19 men), age 26 to 79 years, with normal or corrected to normal visual acuity. Women and men did not differ significantly in age (women: 58 ± 15 years, men: 51 ± 17 years, t(35) = 1.34, ns), handedness (Crovitz and Zener, 1962; 14–32 right handed, 50–70 left handed; women: 20 ± 7, men: 22 ± 12, t(35) = 0.54, ns), education (women: 17 ± 3 years, men: 16 ± 3 years, t(35) = 1.16, ns), or verbal intelligence (NART IQ, women: 115 ± 6, men: 117 ± 5, t(28) = 0.8, ns). All participants received a Structural Clinical Interview for DSM-IV diagnosis (American Psychiatric Association, 1994) by trained clinicians to rule out psychiatric and neurological disease. Written informed consent was obtained from all participants and the study was approved by the local ethics committee. The sample of 37 reflects the exclusion of six subjects, four women who were inattentive while performing the task, i.e., three were talking and one had RTs in the global selective attention task exceeding the group’s mean by more than three standard deviations (SD), and two subjects (1 woman, 1 man) who had ventricle sizes that were more than four standard deviations (SD) from the normal average.

4.2. Stimuli and procedures

Stimuli consisted of the letters E, T, L and F, for example, a global F made out of local Es. Local letters were black; global letters had a light grayish background color to enhance their salience. Local and global letters were presented superimposed on white background. The letters E and T were targets, L and F non-targets. Stimuli were presented at the center of a 21-inch computer screen. Global stimuli were 9.5 cm high and 7 cm wide; local stimuli were 1 cm by 0.5 cm. With a subject-monitor distance of approximately 55 cm, global stimuli subtended ± 5° visual angle vertically and ± 3.5° visual angle horizontally. Local stimuli measured 1° by 0.5° visual angle.

Hierarchical letters were presented in two selective attention blocks of trials and one divided attention block. In the selective attention blocks participants attended to either the global or local spatial scale. In the divided attention block participants attended to both spatial scales simultaneously. Stimuli were the same for each block, only the attention instruction differed.

There were four target conditions: a target letter on the global, the local, both, or neither level (Figure 5). When targets appeared on both spatial scales, they could be a) congruent, i.e., the same letter was presented at the global and local spatial scale, or b) incongruent with different target letters at the two spatial scales, i.e., a global E made out of local Ts.

Participants were instructed to identify target letters (E, T) among distractors (F, L) and pressed a YES button with the index finger of their dominant hand when a target letter appeared at the attended level, and a NO button with the middle finger of the same hand when a non-target appeared at the attended level. Response buttons were the N and M keys on a regular computer keyboard that was aligned to the participants’ body midline. Subjects were instructed to sit about 55 cm from the screen under free viewing conditions (i.e. without a chin rest). Stimuli remained on the screen until the participant pressed the response button ensuring individually optimized processing time for both local and global processing levels. Between trials a blank screen occurred for 30 msec. RTs and errors were collected for each trial. A total of 288 stimuli were presented. Each of the three blocks comprised 32 stimuli with 8 stimuli in each of the four target conditions, and each block was presented three times in the following fixed order: G-L-GL-G-L-GL-G-L-GL. Overall error rates and median RT values for each condition were calculated from all three measurements. Participants performed short practice blocks of trials for each attention condition before testing.

4.3. MRI protocol

The midsagittal area of the corpus callosum and its subareas (genu, body, and splenium) were measured in 37 participants. T1-weighted SPGR images (TR = 26 msec, TE = 5 msec, 30° flip angle, field of view = 24 cm, 256 × 196 matrix, acquired resolution = 0.9 × 1.2 × 2 mm, reconstructed and resampled resolution, 1.0 × 1.0 × 1.0 mm) were acquired on a 1.5T GE Signa MRI system. As previously described (Schulte et al., 2004), measurements were made on a midline slice extracted from the native data after interpolation, alignment, and reslicing. For extraction of the aligned midline slice, the anterior commissure (AC) was identified on the native sagittal slice on which it was best visualized. The sagittal slices were then stacked and interpolated to high isotropic resolution (1 mm). The left/right midline was determined from a coronal reconstruction at the level of the AC. The AC and posterior commissure (PC) were identified on a resliced midline sagittal image. Head tilt angle was defined on a high resolution coronal image and head rotation was defined on a high resolution axial image using an interactive rotating cursor. The resultant landmarks and angles were used to align the volume in a uniform space and orientation anchored on the AC. The midsagittal image was then extracted for semi-automated edge identification of the corpus callosum.

Regional callosal areas were defined geometrically as follows (Sullivan et al., 2002). The corpus callosum silhouette was rotated to a plane parallel to the inferior extremes of the rostrum anteriorly and splenium posteriorly. The midpoint along this plane between the anterior extreme of the genu and posterior extreme of the splenium was used as the center of a circle, and radii were projected at +30 and +150° angles from the plane, thus dividing the corpus callosum into genu + rostrum, body, and splenium.

Because men and women have fundamentally different size brains and substructures (Dekaban and Sadowsky, 1978), we used a linear regression model to correct the size of the corpus callosum for normal variation in intracranial volume (ICV) (Pfefferbaum et al., 1994; Sullivan et al., 2001), based on a sample of 121 healthy men and women who spanned the adult age range of 20 to 81 years (Pfefferbaum et al., 2006); the subjects of the present sample are a subset of the full sample. The corpus callosum and its substructures were then expressed as ICV-corrected standardized Z-scores, with an expected mean of 0 and a standard deviation of 1. In a previous MRI study, which also examined the ventricular system, we found that one man and one woman included in the current report had ventricular volumes that were greater than four standard deviations from the mean of the remaining controls and thus were deemed statistical outliers and excluded from further analyses.

4.4. Statistical analysis

Non-paired Student’s t-tests and χ² tests were used for group comparison (men vs. women) of demographic data. RT analysis of global and local information processing was based on correct responses. Interference and facilitation effects for concurrent (1a,b; 2a) and sequential global-local (1c; 2b) processing were tested: 1) interference from a) incongruent targets at the global and local level (e.g., local E, global T) compared with congruent targets (e.g., local E, global E) (Navon and Norman, 1983), b) targets at the non-attended level that suggest a different response from targets at the attended level, c) switching between target levels in two consecutive trials (global to local, local to global), and 2) facilitation from a) two targets compared with one target at the attended level, and b) target repetition at the same level in two consecutive trials (repetition priming). First, a series of analysis of variance (ANOVAs) used sex as between-subject variable, and the repeated measures component tested for the following effects: precedence, interference between local and global processing level, response conflict and response facilitation, and level repetition and level switching. Secondly, relations of these effects with age and with regional callosal area (genu, body, splenium; Z-scores using regression analyses (Pfefferbaum et al., 1994) corrected for intracranial volume, ICV, for women and men) were tested with two-tailed Pearson correlations, which permitted age to be modeled as a continuous rather than discrete variable. Thirdly, for significant correlations, the predictive value and independent or joint contribution of age and callosal area for the special effects of global-local processing was tested using multiple regression analysis. The alpha level was set to 0.05 for all hypotheses tested (SPSS 10.0).