Elsevier

NeuroImage

Volume 73, June 2013, Pages 225-236
NeuroImage

Comments and Controversies
Teaching an adult brain new tricks: A critical review of evidence for training-dependent structural plasticity in humans

https://doi.org/10.1016/j.neuroimage.2012.03.069Get rights and content

Abstract

A growing number of structural neuroimaging studies have reported significant changes in gray matter density or volume and white matter microstructure in the adult human brain following training. Such reports appear consistent with animal studies of training-dependent structural plasticity showing changes in, for example, dendritic spines. However, given the microscopic nature of these changes in animals and the relatively low spatial resolution of MRI, it is unclear that such changes can be reliably detected in humans. Here, we critically evaluate the robustness of the current evidence in humans, focusing on the specificity, replicability, and the relationship of the reported changes with behavior. We find that limitations of experimental design, statistical methods, and methodological artifacts may underlie many of the reported effects, seriously undermining the evidence for training-dependent structural changes in adult humans. The most robust evidence, showing specificity of structural changes to training, task and brain region, shows changes in anterior hippocampal volume with exercise in elderly participants. We conclude that more compelling evidence and converging data from animal studies is required to substantiate structural changes in the adult human brain with training, especially in the neocortex.

Introduction

The relationship between brain structure and function has gained recent prominence in human neuroimaging. Studies have reported correlations between behavioral performance and localized brain structure (for a recent review, see Kanai and Rees, 2011) and have also identified possible training-dependent changes in structure (e.g. changes in measures of gray matter density or white matter integrity). Evidence for effects of training comes from both cross-sectional studies, comparing different groups of subjects with different experiences (e.g. musicians versus non-musicians (Bengtsson et al., 2005) or taxi versus bus drivers (Maguire et al., 2006)), as well as longitudinal studies, examining the effect of training over time in individuals (for a review see, Draganski and May, 2008, May and Gaser, 2006). However, with cross-sectional studies it is impossible to determine which came first, the structural differences or the experience (May, 2011). Longitudinal training studies provide the strongest evidence for training-dependent changes in brain structure since experience is directly manipulated and the changes are measured within a participant.

Such MRI evidence for adult structural plasticity seems consistent with animal studies of experience-dependent plasticity (Draganski and May, 2008) and based on this apparent convergence, it has been proposed that changes in the MRI signal may reflect changes in axonal myelination, neurogenesis, angiogenesis, dendritic spine motility, glial cell proliferation, and synaptogenesis (Draganski and May, 2008, Scholz et al., 2009). However, while animal studies do suggest that experience-dependent structural plasticity in the adult brain persists throughout the life span (Fu and Zuo, 2011), it is highly constrained. For example, longitudinal in vivo studies suggest that experience does not cause any change in large scale axons and dendrites (Mizrahi and Katz, 2003, Trachtenberg et al., 2002), although some cross-sectional studies have reported changes in glial cells, unmyelinated axons or dendritic length in adult animals exposed to an enriched environment (Juraska et al., 1980, Markham et al., 2009), altered visual input (McBride et al., 2008), or motor tasks (Black et al., 1990, Kleim et al., 2007). There is also no direct evidence for experience-driven increase in axonal myelination in the adult brain (Demerens et al., 1996, Markham et al., 2009). Likewise, experience-dependent angiogenesis in adults has been shown to be specific to exercise (Black et al., 1990, Kleim et al., 2002) (but see Isaacs et al., 1992), and the only undisputed claim regarding neurogenesis in the adult brain is that it is primarily observed in the dentate gyrus of the hippocampal complex and the olfactory bulb (Rakic, 2002).

Overall, the bulk of the evidence from animal studies suggests that experience-dependent structural plasticity is mediated by remodeling of neuronal processes (Lerch et al., 2011, McBride et al., 2008), synaptogenesis (Black et al., 1990, Briones et al., 2004, Knott et al., 2006) or transient changes in dendritic spines (Holtmaat et al., 2006, Trachtenberg et al., 2002, Xu et al., 2009b) and axonal boutons (Stettler et al., 2006, Yamahachi et al., 2009). Importantly, the experience-related increase in structures like dendritic spines is also accompanied by spine elimination resulting in similar total spine densities between the trained and untrained animals after training (Trachtenberg et al., 2002, Xu et al., 2009b). At the systems level, such subtle changes are considered sufficient to remodel patterns of activity in neuronal circuits (Chen and Nedivi, 2010), without inducing large-scale structural alterations in cortical networks. Thus, the evidence from animal studies suggests that the large-scale organization of axons and dendrites is very stable and experience-dependent structural plasticity in the adult brain occurs locally and is transient (for a review see, Holtmaat and Svoboda, 2009).

One of the big advantages of MRI is the capacity to image the whole brain, rather than individual cellular structures as in the case of, for example, 2-photon microscopy. However, given the large-scale stability of structures described above, it is not clear that human MRI, with typically 1 mm3 voxels, can detect the type of microscopic structural changes reported in animal studies. In addition, it is important to note that much of the evidence from animal studies comes from highly invasive or demanding experimental manipulations such as trimming whisker barrels or rearing animals in enriched versus isolated environment, and the animals are motivated by requiring performance to receive food. In comparison, human studies use less intensive and demanding training tasks and some have suggested that a controlled training protocol is not even necessary for inducing structural changes in the adult brain (Bezzola et al., 2011). Finally, we note that the effect size reported in some human studies is very small relative to the size of the voxels. For example, memory training was reported to increase cortical thickness by ~ 0.05 mm (Engvig et al., 2010). Similarly, aerobic exercise was reported to increase hippocampal volume by ~ 0.10 mm3 (Erickson et al., 2011a). Given that such effect sizes are many times smaller than the sampling frequency of the method, these results need to be carefully evaluated and interpreted with caution.

Taking into account these considerations, we conducted a critical review of the evidence from all longitudinal studies of training-dependent structural plasticity in adult humans. There are two key questions. First, how reliable is the evidence for training-dependent changes in MRI measures of brain structure? Second, if there are reliable changes, what do these changes in MRI measures reflect in terms of the biological substrate? Here, we focus primarily on the first question, but will discuss the second issue towards the end of the review. Specifically, in contrast to previous reviews of this literature (Draganski and May, 2008, May, 2011), we focus on the robustness of the experimental design and statistical methods, as well as the limitations of MRI-based structural imaging techniques.

In the first part of the review, we will discuss the different methods used to measure human adult structural plasticity and briefly survey the extant literature. In the central part of the review, we will evaluate the reported findings in terms of specificity, replicability and correlation with behavior. Finally, we will consider in more detail the inherent limitations of MRI measures of structure and the relationship between MRI measures and the biological substrate.

Section snippets

Measuring training-dependent structural changes

In total, we identified 20 research articles (see Table 1) that satisfy the following inclusion criteria: (a) the studies involved healthy adults (mean age > 18). (b) A longitudinal design was employed and participants were scanned before and after training in a specific task. (c) MRI-based techniques were used to measure structural changes (Fig. 1). The training tasks employed in these studies range from visuomotor tasks such as, juggling (for e.g., Draganski et al., 2004), golf (Bezzola et al.,

Specificity

The level of specificity of any apparent training effect is critical for its interpretation. At a basic level, it is important to demonstrate that any changes are specific to training and are not observed in untrained subjects. However, trained and untrained subjects are not well matched in terms of their overall experience and therefore, demonstration of specificity to a particular task provides more compelling evidence. Further, finding specificity to a particular task can give important

Replicability

Three-ball juggling has been used as the training task in four separate studies (see Table 1) allowing us to evaluate the replicability, of the structural changes reported (Fig. 3). In the first of these studies, Draganski et al. (2004) reported changes in GMD bilaterally in the middle temporal region (visual motion area hMT/V5 — although hMT was not functionally localized in this or later studies) and near the left posterior intra-parietal sulcus (IPS). Further, the increases observed with

Correlation with behavior

Individual variations in GMD and white matter integrity have been reported to account for variance in behavioral measures (for a review, see Kanai and Rees, 2011). If the reported structural changes are the direct result of training, it seems reasonable to suppose that the change in structure should correlate with some measure of training behavior. Such a correlation is not necessary to conclude training-dependent structural changes but would significantly bolster support for this conclusion,

Robustness of evidence

So far our review of the strength of the existing evidence for training-dependent structural plasticity in adult humans reveals a number of limitations. In many cases, the statistical evidence does not appear to be robust and the evidence for replication is weak. Out of 20 studies, only one (Erickson et al., 2011a) demonstrates effects that are specific to training, task and brain region, with a significant correlation with behavioral performance (but see, Coen et al., 2011, Erickson et al.,

Challenges in MRI based structural imaging

As noted earlier, unlike the structural imaging techniques used in animal studies, MRI-based techniques have relatively poor spatial resolution. Moreover, in both techniques the raw data undergoes several stages of processing to provide a measure related to the underlying biological structure. While these procedures are employed in order to reduce the effect of various sources of noise and to improve statistical inference, they make many assumptions and may introduce specific biases into the

Interpreting the evidence from MRI-based structural imaging

So far we have considered the reliability of evidence for training-dependent changes in MRI measures of brain structure. However, a second critical question is what any changes in human MRI measures, such as cortical thickness, gray matter density or fractional anisotropy, might reflect in terms of the biological substrate and how they relate to the animal literature. Specifically, the T1-weighted structural images that are used as the raw data for morphometric analysis such as cortical

Conclusion

Based on our review of the literature and the limitations of MRI-based measures of structure, we conclude that the current literature on training-dependent plasticity in adult humans does not provide unequivocal evidence for training-dependent structural changes and more rigorous experimentation and statistical testing is required. Of the 20 studies we reviewed here, only one (Erickson et al., 2011a) provides strong evidence for effects that are specific both to the training task and to

Acknowledgments

This work was supported by the Intramural Research Program of NIMH. Support for this work also included funding from Department of Defense in the Center for Neuroscience and Regenerative Medicine. We thank Alan Koretsky, Sean Marrett, Alex Martin, Carlo Pierpaoli, Adam Thomas, Marta Ceko, and members of the Laboratory of Brain and Cognition, National Institute of Mental Health, for their helpful comments and discussion. Special thanks to Ziad Saad for the help in producing Fig. 3.

References (114)

  • C.W. Cotman et al.

    Exercise: a behavioral intervention to enhance brain health and plasticity

    Trends Neurosci.

    (2002)
  • W. Crum et al.

    Zen and the art of medical image registration: correspondence, homology, and quality

    NeuroImage

    (2003)
  • C. Davatzikos

    Why voxel-based morphometric analysis should be used with great caution when characterizing group differences

    NeuroImage

    (2004)
  • E. Dayan et al.

    Neuroplasticity subserving motor skill learning

    Neuron

    (2011)
  • B. Draganski et al.

    Training-induced structural changes in the adult human brain

    Behav. Brain Res.

    (2008)
  • B. Draganski et al.

    Regional specificity of MRI contrast parameter changes in normal ageing revealed by voxel-based quantification (VBQ)

    NeuroImage

    (2011)
  • A. Engvig et al.

    Effects of memory training on cortical thickness in the elderly

    NeuroImage

    (2010)
  • S. Eriksson et al.

    Quantitative grey matter histological measures do not correlate with grey matter probability values from in vivo MRI in the temporal lobe

    J. Neurosci. Methods

    (2009)
  • K. Friston et al.

    A critique of functional localisers

    NeuroImage

    (2006)
  • M. Fu et al.

    Experience-dependent structural plasticity in the cortex

    Trends Neurosci.

    (2011)
  • S. Hayasaka et al.

    Validating cluster size inference: random field and permutation methods

    NeuroImage

    (2003)
  • S. Hayasaka et al.

    Nonstationary cluster-size inference with random field and permutation methods

    NeuroImage

    (2004)
  • D.K. Jones et al.

    The effect of filter size on VBM analyses of DT-MRI data

    NeuroImage

    (2005)
  • J.M. Juraska et al.

    Plasticity in adult rat visual cortex: an examination of several cell populations after differential rearing

    Behav. Neural. Biol.

    (1980)
  • J.A. Kleim et al.

    Exercise induces angiogenesis but does not alter movement representations within rat motor cortex

    Brain Res.

    (2002)
  • J.A. Kleim et al.

    Motor learning induces astrocytic hypertrophy in the cerebellar cortex

    Behav. Brain Res.

    (2007)
  • A. Klein et al.

    Evaluation of 14 nonlinear deformation algorithms applied to human brain MRI registration

    NeuroImage

    (2009)
  • N.A. Lazar et al.

    Combining brains: a survey of methods for statistical pooling of information

    NeuroImage

    (2002)
  • J.P. Lerch et al.

    Cortical thickness analysis examined through power analysis and a population simulation

    NeuroImage

    (2005)
  • J.P. Lerch et al.

    Maze training in mice induces MRI detectable brain shape changes specific to the type of learning

    NeuroImage

    (2011)
  • M. Lövdén et al.

    Experience-dependent plasticity of white-matter microstructure extends into old age

    Neuropsychologia

    (2010)
  • J.F. Mangin et al.

    Distortion correction and robust tensor estimation for MR diffusion imaging

    Med. Image Anal.

    (2002)
  • S. Marenco et al.

    Regional distribution of measurement error in diffusion tensor imaging

    Psychiatry Res. Neuroimaging

    (2006)
  • J.A. Markham et al.

    Myelination of the corpus callosum in male and female rats following complex environment housing during adulthood

    Brain Res.

    (2009)
  • A. May

    Experience-dependent structural plasticity in the adult human brain

    Trends Cogn. Sci.

    (2011)
  • A. May et al.

    Response to Thomas and Baker: the structural adaptation of the brain to training

    Trends Cogn. Sci.

    (2012)
  • R. Poldrack

    Imaging brain plasticity: conceptual and methodological issues — a theoretical review

    NeuroImage

    (2000)
  • G.R. Ridgway et al.

    Ten simple rules for reporting voxel-based morphometry studies

    NeuroImage

    (2008)
  • G. Salimi-Khorshidi et al.

    Meta-analysis of neuroimaging data: a comparison of image-based and coordinate-based pooling of studies

    NeuroImage

    (2009)
  • C. Salmond et al.

    Distributional assumptions in voxel-based morphometry

    NeuroImage

    (2002)
  • R. Saxe et al.

    Divide and conquer: a defense of functional localizers

    NeuroImage

    (2006)
  • T. Schmidt-Wilcke et al.

    Distinct patterns of functional and structural neuroplasticity associated with learning Morse code

    NeuroImage

    (2010)
  • M. Silver et al.

    False positives in neuroimaging genetics using voxel-based morphometry data

    NeuroImage

    (2011)
  • S.M. Smith et al.

    Accurate, robust, and automated longitudinal and cross-sectional brain change analysis

    NeuroImage

    (2002)
  • D.D. Stettler et al.

    Axons and synaptic boutons are highly dynamic in adult visual cortex

    Neuron

    (2006)
  • C. Thomas et al.

    Remodeling human cortex through training: comment on May

    Trends Cogn. Sci.

    (2012)
  • C.I. Baker et al.

    Does the fusiform face area contain subregions highly selective for nonfaces?

    Nat. Neurosci.

    (2007)
  • P.J. Basser et al.

    Diffusion tensor MRI: theory, experimental design and data analysis — a technical review

    NMR Biomed.

    (2002)
  • C. Beaulieu

    The basis of anisotropic water diffusion in the nervous system — a technical review

    NMR Biomed.

    (2002)
  • C. Beaulieu
  • Cited by (169)

    • White matter

      2021, Encyclopedia of Behavioral Neuroscience: Second Edition
    View all citing articles on Scopus

    No conflicts of interest.

    View full text