Reproducibility of Complex Functional Magnetic Resonance Imaging Effects.

Abstract

To the Editor: Recent research suggests low reproducibility of experimental psychological studies.1 Because there is a great variety of paradigms, models, and analytical approaches, this may particularly apply to studies using functional magnetic resonance imaging (fMRI). Consequently, confirmation of previous results using different tasks, designs, or statistical methods is desirable and necessary to verify specific and complex effects. However, replications are rare, probably because reporting similar results reduces the likelihood of a high-impact publication. In the last decade, several working groups have focused on neurocognitive changes associated with normal aging. Different factors such as performance level and task load strongly modulate these changes. As a result, findings are inconsistent depending on which of these factors are considered in study design. Greater prefrontal activation or bilaterality in older adults together with age-related performance decline, for example, argue for neural inefficiency, less regional specificity, or failed compensation, whereas overactivation or greater bilaterality at a steady performance level may be signs of successful compensation. By contrast, less prefrontal activation in older adults associated with lower performance accuracy can be interpreted as neural dysfunction, whereas less prefrontal activation associated with steady or better performance indicates greater efficiency. Moreover, older adults have greater bilateral activation at low task load as a sign of compensation and lower activation at high task load as a sign of exhausted neural resources.2 One of the few studies that have included both performance and task load in a single analysis revealed activation differences between younger high-performing, younger low-performing, older high-performing, and older low-performing subjects. Most importantly, the results point toward a more-efficient or ‘youth-like’ load-dependent up-regulation of the spatial working memory network in older high-performing subjects.3 Recent results4 confirmed these effects, although a different paradigm, a different design, and a different methodological approach were used (e.g., serial vs parallel stimulus presentation; retrieval vs recognition; load levels 4, 5, and 6 vs 1, 3, and 7; different region of interest masks and analysis of variance designs). Nevertheless, particularly for the left dorsolateral prefrontal cortex, activation patterns modulated by performance level and task load are surprisingly similar, which provides further evidence of the validity of these effects (compare left Figure 5A in3 with bottom Figure 4 in4). In addition, it appears that cerebral up-regulation of younger low-performing subjects plateaus at moderate load levels, whereas the neural resources of older low-performing subjects seem to be exhausted already at low task load. A better understanding of these mechanisms is an important prerequisite for interpretation of functional changes in the aging brain, which makes their replication indispensable. The studies described show that even complex fMRI effects are reproducible and provide a good example of the validity of experimental psychological research. Conflict of Interest: The authors declare no conflicts of interest. Author Contributions: All authors contributed equally to this letter. Sponsor's Role: There was no external sponsorship.