Reply to: Assessing prospective memory beyond experimental tasks

Two statements from my recent Review on prospective memory (PM) impairment (Henry, J. D. Prospective memory impairment in neurological disorders: implications and management. Nat. Rev. Neurol. https://doi.org/10.1038/s41582-021-00472-1 (2021))¹ are highlighted as warranting discussion by Hainselin et al. (Hainselin, M., Gounden, Y. & Blondelle, G. Assessing prospective memory beyond experimental tasks. Nat. Rev. Neurol. https://doi.org/10.1038/s41582-021-00499-4 (2021))², namely, “self-report measures often correlate weakly with objective assessments” and “single-item PM tests have lower reliability and sensitivity than clinical batteries.” Hainselin et al. argue that “recent literature reviews have highlighted the flaws of clinical batteries and emphasized the relevance of questionnaires and single-item PM tests.”

As I discussed¹, self-report scales provide valuable insights into a patient’s own perspective, but, because they often correlate weakly with objective assessments, they should supplement rather than replace a formal behavioural assessment. This viewpoint aligns completely with the following conclusions from the review that Hainselin et al.² themselves cite³: “this review found that self- and informant-report measures have relatively weak relationships with performance-based measures of PM. Some limited evidence of self-report and informant-report measures being able to detect PM impairments and monitor intervention outcomes is reported. As such, these measures are most suitable for the measurement of individuals’ concerns and beliefs about their PM ability and the impact of PM failures on their lives rather than measures of PM ability itself.”

With respect to objective PM measures being imperfect, this is true: PM is an extraordinarily complex cognitive ability to measure. I particularly agree about the need for increased ecological validity⁴ and greater testing and validation in non-WEIRD (Western, educated, industrialized, rich and democratic) populations, not only for PM but also for many other neurocognitive abilities, such as social cognition⁵. However, many of the other concerns raised do not apply to the four performance-based PM tasks referred to in my Review¹. As detailed in my Review, three of the tasks are available in three or more languages and formal norms are available for two tasks. Qualitative scoring is minimal for all four tasks and, in one case, has been eliminated via computer automation. All four tasks have good validity in terms of sensitivity to PM impairment (for three, sensitivity has been shown in eight or more distinct clinical populations).

With respect to the value of single-item tools, in the broader neurocognitive literature, single-item assessments are rarely recommended and are particularly problematic for the assessment of complex neurocognitive abilities such as PM. A key point of my Review was that, because PM impairment can reflect a breakdown in many distinct neurocognitive resources, consideration of patterns of performance across different PM tasks is crucial¹ and is not possible with a single-item probe.

In addition, basing clinical decision-making on the results of an individual test item is problematic. Indeed, one of the two papers that Hainselin et al.² cite as providing support for the value of the Envelope Task⁶ cautioned that “sensitivity was moderate at most (64.3%), which likely relates to its limited scale, and suggests that many cases of amnestic mild cognitive impairment could be missed if this test were interpreted in isolation.”

In summary, it seems prudent to only consider single-item tests as the most basic of screening tools or as a method for illustrating to patients or their families how PM failures can cause problems completing everyday tasks.