Tasks that are used to examine pseudoneglect are also used to assess hemispatial neglect in clinical populations, and research examining populations with neglect often used control subjects who are matched in age to patients who are typically in or beyond their fifth decade of life. Understanding the normal variability demonstrated on visuospatial tasks by older adults can assist researchers and clinicians in interpreting the findings observed in clinical populations (e.g., patients with hemispatial neglect). The aim of the current systematic review is to integrate the available research on pseudoneglect in late adulthood to discuss the association between age and a bias to the left hemispace. Synthesizing the literature on age effects will contribute to an understanding of the normal variability in pseudoneglect demonstrated by neurologically healthy older adults.
Discussion
The aim of the systematic review was to aggregate and summarize age-related differences in performance on tasks used to examine pseudoneglect. Following a systematic search for relevant studies, multiple studies were identified that employed the line bisection task, and a smaller number of studies utilized the landmark, greyscales, tactile rod bisection, and lateralized visual detection tasks. Together, the literature retrieved was characterized by inconsistent results and large variability in study design. Unsurprisingly, this conclusion is identical to the finding reported in Jewell and McCourt’s (
2000) qualitative review of the line bisection literature. Even when studies employed identical tasks, they varied in methods (e.g., hand used, direction of scanning), stimuli (e.g., stimulus length, number of stimuli viewed), and approach to comparing participants (e.g., gender, handedness), including structure of age groups (e.g., number of age groups, age range within groups). These differences make it difficult to assess the degree to which performance is influenced by age; thus, it is premature to draw conclusions based on the literature included in the review. However, when comparing the identified studies that examined age-related differences in pseudoneglect a number of observations were noteworthy.
All of the studies included in the review used a cross-sectional design. The analysis of cross-sectional samples varying in age is consistent with the research paradigm in gerontology that has been predominately used to understand cognitive aging (Hofer et al.
2002). However, researchers have questioned the utility of using cross-sectional studies to understand age-related changes, and have argued that understanding aging also requires analysis of change within individuals (Hofer et al.
2002). Relying solely on cross-sectional research to understand pseudoneglect across the life span may be misleading and is unlikely to provide an accurate understanding of longitudinal change or the effect of chronological age. Cross-sectional data has been found to provide unreliable estimates of age-related cognitive decline by conflating the effect of age with cohort effects (Singh-Manoux et al.
2012). Further, research examining cognitive aging has commonly reported discrepancies between cross-sectional and longitudinal age trends with between-person cross-sectional comparisons reporting declines in functioning beginning in early adulthood, whereas within-person longitudinal comparisons report stability or increases in cognitive performance (Salthouse
2010). In an area of research that is dominated by cross-sectional associations between age and pseudoneglect, longitudinal data that are adequately powered are essential for drawing conclusions regarding change with chronological age. To fully understand pseudoneglect across adulthood, future research could benefit from basing conclusions on results derived from multiple methods of data collection and analysis (Salthouse
2011).
Further, when examining cross-sectional differences, the age ranges studied varied substantially. The majority of studies included in the review used a modal “extreme age group design” (Marsiske and Margrett
2006, pp. 320) and categorized participants into younger and older adult groups, with the age range within the older adult age group spanning 20 to 30 years (e.g., 60–80 years or older). Comparing extreme groups of younger and older adults is problematic as the variance associated with middle-aged adults is omitted and inflates estimates of age-related differences. Further, because changes in cognitive functioning often occurs continuously across adulthood, results based on lateral biases observed over a large period of older age (e.g., 20–30 years) could be misleading with regard to the origin of age-related differences and whether the identified age relations are linear. Fewer studies included a middle age group, and only three studies included in the review categorized age groups with smaller age ranges (e.g., 10-year cohorts; Beste et al.
2006; Friedrich et al.
2016; Varnava and Halligan
2007). These three studies reported an enhanced leftward bias with age (Friedrich et al.
2016), particularly as demonstrated by women (Varnava and Halligan
2007), or reported comparisons between age groups that did not reach levels of significance (Beste et al.
2006).
The use of broad age categories and cross-sectional design may be contributing to the variability of perceptual biases observed in older adults within and between the various tasks used to examine pseudoneglect, and may be inhibiting researchers’ ability to understand
when changes occur. Improving design by examining age groups with smaller age ranges, or using longitudinal methods, is critical, as the extent of changes in cognitive performance may vary considerably over a large age range (e.g., 20 to 30-year span) in older age. Research on the course of intellectual abilities, including spatial orientation, over the adult lifespan has revealed that performance plateaus after a peak in young adulthood until the late 50’s or early 60’s, and then declines at a slow pace until the late 70’s, when decline is often accelerated (Schaie
1994). Specifically, research over 35 years (six testing cycles) within the Seattle Longitudinal Study showed that decline in cognitive abilities is not reliably confirmed prior to age 60, and fewer than half of the participants showed reliable decrements at age 74; however, by age 81, most abilities decline by one standard deviation (Schaie
1993,
1994). On this basis, large age ranges spanning from 60 to the late 80’s, as typically seen in studies of pseudoneglect, are likely insufficient, and may be resulting in large within group differences leading to the reporting of central or attenuated leftward biases. Thus, the conclusion that pseudoneglect becomes rightward with age may be invalid. Of the studies included in the review, only Friedrich et al. (
2016) categorized participants in 10-year age cohorts and examined adults over 80 years of age as a separate age group. Interestingly, of the seven age groups examined by Friedrich et al. (
2016), only the oldest age group (80–89 year olds) demonstrated an asymmetry score that was significantly different from the youngest age group (18–29 year olds).
Further, when studies differ in comparison groups (e.g., comparing a sample of younger adults to a sample of older adults, comparing three or more age groups) and focus on different age ranges, it may not be meaningful to treat the results obtained as equally comparable. For example, differences that have been reported to occur at approximately 60 years of age (mean age of participants was 61.6 years; Fukatsu et al.
1990) may not involve the same mechanisms as age-related differences that have been reported to occur at 75 years of age (mean age of older participants was 74.6 years; De Agostini et al.
1999). Using smaller age ranges within age groups will likely assist researchers in observing age-related differences, and differentiating an age at which there is a reliably detectable change in pseudoneglect. Understanding when age-related differences begin to occur (i.e., mid-life versus very-late life) and examining the specific age groups identified will assist in enhancing the value and relevance of the research.
Furthermore, a consistent categorization of older adults and specification of whether participants are neurologically healthy in the studies reviewed was limited. Large individual difference in cognitive performance and rate of change observed with aging may also be contributing to the variability in findings. The cognitive status of older adults has been found to have extensive heterogeneity and, despite accounting for clinical diagnoses (e.g., normal cognitive function, MCI, dementia), rates of change can vary from a decline of 0.3 SD per year to improvements of 0.1 SD per year (Mungas et al.
2010). The cognitive status of older individuals is complex and influenced by many variables in addition to age, including, but not limited to, brain injury and disease, mental health, health status, and exposure to substances and medications (Mungas et al.
2010). The multiple deleterious and protective factors that influence the variance in cognitive function and rate of change with age could also be influencing the inconsistent findings revealed in this systematic review.
The majority of studies included in the review examined adults over 60 years of age and did not screen for symptoms of neuropathology, such as mild cognitive impairment or dementia. Of the 32 studies, six screened for mild cognitive impairment and examined whether younger and older age groups differed in general cognitive performance. When cognitive performance was assessed, older adults demonstrated accurate bisections (Barrett and Craver-Lemley
2008; Chieffi et al.
2014; Mendez et al.
1997) or demonstrated leftward spatial bias that did not differ from younger adults (Brooks et al.
2016; Chen et al.
2011 (female participants); McPherron
2015), except for Chen et al. (
2011) who identified an interaction between sex and age with only male participants demonstrating rightward bisection biases with age. Given that cognitive screening was employed in a limited number of studies included in the systematic review, it is difficult to determine the presence of neuropathology and whether age-related differences are related to
healthy aging. Future research would benefit from examining lateral biases in healthy older adult populations by incorporating measures that screen for symptoms mild cognitive impairment. Further, it would be useful to compare the lateral biases of older adults with and without symptoms of cognitive impairment to examine the effects of neuropathology on age-related differences in pseudoneglect and understand the continuum of normal and pathological aging. Of the 32 titles included in the systematic review, only two titles compared the performance of older adults with and without symptoms of pathological aging (McPherron
2015; Mendez et a., 1997) and comparisons did not reach statistical significance.
Individual characteristics in addition to age, such as gender (see Jewell and McCourt
2000, for review) have also been investigated to understand the association between gender and lateral perceptual biases. Of specific interest, a number of manual line bisection studies that surfaced in this review reported sex-differences in age-related differences on bisection performance. Age effects appeared to be stronger in males, as men typically demonstrated an attenuated leftward bias or rightward bias, whereas women demonstrated a leftward bias comparable in magnitude to younger participants (Barrett and Craver-Lemley
2008; Chen et al.
2011; Pierce
2000; Varnava and Halligan
2007). However, Beste et al. (
2006) reported discrepant results, as men in each age group bisected to the left of true center when using their left hand, whereas women in all age groups bisected to left of true center except for women 50 to 59 years of age. In contrast, tasks that reduce the influence of motor cuing (e.g., landmark and greyscales task) failed to find differences in the magnitude of pseudoneglect between males and females (Benwell et al.
2014; Friedrich et al.
2016; Learmonth et al.
2017; Schmitz and Peigneux
2011; Schmitz et al.
2013). Consistent with a hypothesis proposed by Benwell et al. (
2014), gender- specific aging effects may be influenced by non-perceptual factors, such as motor cueing. Future research that accounts, or controls, for stimulus factors and experimental methods will assist in deconstructing gender specific aging effects.
Heterogeneity within tasks with regard to differences in the stimuli used and method used to calculate the dependent variable decreases internal comparability. For example, studies that used the line bisection task differed in the size and the number of stimuli presented to participants, which also influenced the number of trials used to calculate lateral biases. In the studies included in the review, stimulus length varied from 20 to 400 mm in length, the number of different lengths presented varied from two to 13 different lengths, and presentation of the lines varied from presentation of a single line to multiple lines on a page. Studies that used the line bisection task also differed with regard to which hand participants were instructed to use. Some studies specified using the right, left, or both hands (i.e., bimanual), whereas hand use was not specified in other studies. Studies that used similar tasks also differed in how perceptual biases were calculated. For example, studies that used the landmark task calculated a leftward perceptual bias as the percentage of left longer/right shorter responses for evenly bisected lines (Schmitz and Peigneux
2011; Schmitz et al.
2013), others calculated a leftward response based on the number of leftward choices participants made when asked which end of the line they thought the transection was closest to (Harvey et al.
2000), and still others used a cumulative logistic psychometric function (i.e., a measure of the precision of midpoint judgments) and a point of subjective equality (i.e., the perceived midpoint of the line) for unevenly bisected lines (Benwell et al.
2014; Learmonth et al.
2017). Differences in the duration of the stimulus presentation and examination of response time also varied. Stimulus presentation ranged from 150 ms to 1000 ms, and instructions regarding responses varied from no time limits to instructions that emphasized responding as quickly as possible. Such differences in task instructions, administration, and scoring may affect participants performance. Standardization of stimuli and methods of analyses will assist in internal comparability between studies that use similar tasks. Standardization is a requirement for basic experimental control to minimize biased results and the influence of extraneous variables on participants’ performance (Fischer and Milfont
2010). To further understand age-related differences in pseudoneglect within each task discussed, standardization of instructions, administration, and scoring will be imperative. Similar to the administration of standardized psychometric testing, researchers examining pseudoneglect may benefit from the development of and standardization of visuospatial tasks. Standardized stimuli, administration, and scoring would assist in enhancing the internal comparability between studies, and the validity and reliability of the results obtained from testing.
Furthermore, the heterogeneity among the types of tasks used to assess pseudoneglect, including differences in task demands, decreases comparability between tasks. Previous research has failed to find evidence for the inter-task reliability of the line bisection, landmark, greyscales, and lateralized visual detection tasks, and this is proposed to result from differences in task demands (Learmonth et al.
2015a; Rueckert et al.
2002). The line bisection and landmark task have been considered to rely on
global size judgment as both tasks involve assessing the midpoint along a horizontal line, whereas the greyscales task involves a
luminance judgment and the lateral visual detection task involves a
stimulus detection (Learmonth et al.
2015a). Lateralized spatial biases may be task-dependent and assumptions of equivalence in future reviews may be counterproductive.
However, it is also conceivable that improvements in research design, including smaller age ranges, screening for cognitive impairment, and standardization of tasks, may improve internal comparability, but may not improve reported inconsistencies. If the present inconsistencies in research examining age-related changes in pseudoneglect prove robust to improvements in research methodology, the field may find it necessary to acknowledge this pattern within the results and critically consider the validity of the findings. As such, given the variability in the conclusions reported by the studies included in the current review, the visuospatial tasks examined may not provide valid or reliable estimates of age-related changes in cognitive functioning. Specifically, the tasks included in the current review may not be sensitive enough to reliably differentiate the magnitude of pseudoneglect demonstrated by younger and older adults. Rather, the tasks may provide the greatest utility to clinicians when examining patients with brain injuries to assess for larger systematic biases, such as hemispatial neglect.
Limitations
Arguably, a main limitation of this systematic review is the search strategy and eligibility criteria employed. One might have used additional keys words or subject headings to identify the “situation” (i.e., pseudoneglect). However, search terms used in the current study were identified in collaboration with a university librarian and content expert (LE) to enhance the identification of relevant articles. One might have also employed an alternative search strategy. For example, one could have conducted an additional search for studies involving pseudoneglect, regardless of the “population” (i.e., older adults), and subsequently screened for studies examining participants over the age of 60. Following the search employed in the current study and a search using only the “situation”, the studies selected from both searches could be compared to each other to ensure that the search was inclusive. Although this approach was not employed in the current study, forward and backward searching was conducted to enhance the likelihood that the search was exhaustive. Furthermore, with regard to the eligibility criteria, a limitation of the study is that the titles and abstracts were screened and the relevant articles were examined for eligibility by one author (TF); however, if there was doubt regarding whether to include or exclude an article during abstract screening, the article was included for full-text screening.
Another potential limitation of this review is the chance of publication bias. Overall, a large number of studies retrieved and included in the review (10) reported statistical comparisons between age groups that did not reach significance. Thus, there is a high chance that a number of other completed studies may not have been published due to inconclusive results. In an attempt to minimize this bias, grey literature was included in the review. Further, a large number of search terms that are synonymous with pseudoneglect were used and an inclusive inclusion criterion was used to screen for articles to allow for a broader and comprehensive overview of the research that included grey literature.