Learning Specificity and Segmentation Strategies: Misconceptions Regarding Computerized‐Cognitive Training Programs

To the Editor: The debate regarding the efficacy of computerized-cognitive training programs (CCTPs) has escalated over the last few years, with groups of researchers publishing “consensus” statements with opposing views on the topic. Much of the debate has been oriented toward determining the clinical significance of the statistically significant benefits of CCTPs. Lampit and colleagues recently published an article in the Controversies in Aging section of this journal in which they defended the “clear evidence” that CCTPs are effective in maintaining cognitive health in older adults.1 Ratner and Atkinson challenged this claim in a reply article, questioning the meaning of the word “beneficial” in such a context, again approaching the debate from the clinical versus statistical significance angle. The present letter aims at reorienting the discussion toward fundamental learning principles and especially challenges Lampit and colleagues' claim that concerns about transfer of the benefits of CCTPs to real life “are largely unfounded.” Such concerns are actually founded on more than a century of fundamental research showing that learning transfer is a capricious property and that, for it to happen, some specific criteria—that are still not well understood—must be met. This conclusion can be drawn from studies encompassing a broad range of brain functions, from primary sensory processing to complex cognitive tasks. The specificity of learning theory–stating that transfer is possible only when high correspondence between learning and transfer conditions is achieved—has roots more than a century deep.2 Although originally developed in motor learning, it was shown to be a ubiquitous property in perceptual learning as well, strongly established in vision, hearing, and time perception.3 Minor changes to a simple sensory task can prevent any transfer of overpracticed behaviors to another task or stimulus. It was also applied to broader cognitive processes through the encoding specificity principle of memory.4 A classic experiment5 illustrates how crystallized cognitive abilities (in this case, the mnesic skills of elite chess players) in a specific task do not translate to a similar task with minor modifications (plausible vs implausible chess situations). A review of the literature on automobile driving in elderly adults also concluded that learning specificity applied in this more ecological set-up.6 It found that training programs targeting complex sequences of actions, closest to real driving, such as driving in a simulator, led to better driving improvements than in-class programs and basic sensory training. The chunking or segmentation approach of cognition subtending CCTPs is also a source of concern as far as potential generalization in real life. Processing approaches emphasize the importance of the relationship between the different processing systems involved in decoding the environment. This approach has largely replaced structural theories, which depict cognition as an ensemble of isolated structures, notably because of increasing knowledge about the neurophysiology of the brain. A previous study7 showed that targeting higher cognitive strategies instead of isolated cognitive functions might lead to better real-life improvements because it allows communication between the different brain regions involved in various perceptual and cognitive processes, favoring the integration of the dynamic component of thoughts and behaviors. This can be paralleled to motor learning studies showing the importance of learning the dynamics between the portions of a sequence as opposed to training isolated segments.8 The discrepancy between current knowledge about neuroplasticity and behavioral data showing learning inflexibility has been subject to much theoretical gymnastics over the last few years (see 9 for a review). Although at some point transfer was thought to be outright inexistent, it has been argued that some forms of transfer can occur in specific paradigms through repetitive training.10 A few recent studies on learning transfer show promise, but they surely do not eclipse the overwhelming empirical demonstrations of transfer fragility and have not brought enough substance to dissipate scientists' concerns about the rationale of CCTPs. All the more so because contradictive data show that overpracticing a behavior might reduce its potential for generalization.9 Functioning in the everyday life rarely involves isolated and simplified behaviors but instead involves complex sequences of actions, with indissociable sensory, cognitive, motor, motivational, and alertness components whose integration into a comprehensive theory of learning is currently lacking.11 There is no doubt among scientists that the human brain has lifelong plasticity (http://www.cognitivetrainingdata.org/), but the optimal way to take advantage of this property in the fight against cognitive decline has yet to be elucidated. Perfecting CCTPs might represent one avenue, but one has to be cautious about putting all eggs in the same basket and consider other validated options. Conflict of Interest: The authors declare no conflict of interest. Author Contributions: Both authors contributed equally to the writing of the letter. Sponsor's Role: None.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.