Introduction

Non-pharmacological interventions for people at risk of cognitive decline and diagnosed with neurodegenerative diseases (such as Alzheimer’s disease [AD]) include numerous different approaches. Interventions range from yoga (Grzenda et al. 2024) and dance (Ruiz-Muelle and López-Rodríguez 2019) to traditional computer- or paper-based interventions (Georgopoulou et al. 2023). Current studies have shown positive effects of cognitive training in the AD group – it can help stabilise, slow, or even slightly reverse cognitive decline (Georgopoulou et al. 2023; Nousia et al. 2018). Cognitive training can also be a complementary, holistic form of support for people with cognitive decline without dementia. It has no side effects, can enhance the impact of pharmacotherapy, and can be performed by volunteers or educated/trained family members (Livingston et al. 2020).

One approach that should be implemented in all non-pharmacological interventions for people with cognitive impairment is the errorless learning (EL) approach (Clare and Jones 2008; Fillingham et al. 2006; Fillingham et al. 2003). It is based on the 1963 assumption that, in some situations, incorrect behaviour can become depreciating (Clare and Jones 2008). Mistakes in recall and overcomplicated tasks can reinforce the experience of failure, ultimately increasing task difficulty and reducing engagement (Śmigórska et al. 2019). EL underlines the necessity of adapting tasks and is complementary to the latest findings – positive experiences and performance success increase neuroplasticity (El-Mansoury et al. 2025).

This study aimed to evaluate the effectiveness of a structured cognitive intervention based on EL for Polish patients with AD, which, if effective, could then be made available free of charge to interested rehabilitation specialists. Two main hypotheses of the study were:

H.1. The cognitive interventions carried out result in improved performance on cognitive tests in the AD group.

H.1.1. The improvement in cognitive functioning will be sustained across repeated assessments of cognitive function after 12 weeks of training.

The study’s clinical endpoint consisted of a comprehensive neuropsychological evaluation administered at three time points, the last occurring six months after the cessation of the structured intervention.

Material and methods

A longitudinal, repeated-measures paradigm is a recognised methodological approach in dementia research. Establishing a strictly matched control group presents significant logistic and ethical challenges, as comorbidities and varying rates of neurodegenerative progression often confound inter-group comparisons. Consequently, this study used an intra-subject control design, where baseline performance served as the primary reference point for evaluating the intervention’s longitudinal impact. The difficulty of achieving a homogeneous control cohort that accurately reflects the complex clinical profiles of AD patients often necessitates the quasi-experimental framework.

The quasi-experiment presented in this article consisted of completing tasks during twelve weeks of the intervention and four cognitive assessments (before enrolment, immediately after training, three months after training, and six months after training). By treating each participant as their own “control”, the study minimised the impact of baseline variance, providing preliminary insights into the durability of cognitive maintenance. Table 1 presents the methods used to assess cognitive function in the study.

Population of the study

Nine patients (5 females, 4 males) of Caucasian origin, aged 65 to 79 (mean = 73.33, SD = 4.15), diagnosed with late onset AD (based on the criteria defined by the National Institute on Aging and the Alzheimer’s Association working groups (Knopman et al. 2018)) with Cognitive Reserve Index questionnaire (CRIq) scores in the range 90-150 (mean = 111.56, SD = 17.4). All participants gave informed consent, were not involved in other cognitive studies, and were recruited from the Neurology Outpatient Clinic at the University Hospital in Kraków.

Duration and location of the study

The study was conducted between September 2023 and May 2025. The protocol was implemented in a controlled setting, requiring participants to attend two weekly, hour-long sessions for 12 weeks at a location outside their home (the library of the Neurology Outpatient Clinic at the University Hospital in Kraków).

Materials used in the study

Participants completed tailored cognitive handouts prepared for this project. The tasks were divided into five cognitive areas: memory, language, attention, visuo-spatial and executive functions. The task’s difficulty increased and depended on the correct completion of the previous level, in accordance with the principle of errorless learning.

Statistical analysis

Statistical analysis was performed using JASP software [12]. Due to the small sample size, non-parametric methods were used – the paired samples Wilcoxon signed-rank test.

The study received a positive opinion from the Bioethics Committee of the Jagiellonian University Medical College, number 1072.6120.306.2022.

Results

Despite the small sample size, the results showed a normal distribution for most ACE-III subscales (excluding the visuospatial, attention II, and fluency III measures). Comprehensive results, including subdomain analyses, are shown in Figures 1 and 2 and Table 2. More detailed analysis can be found in the supplementary materials (Supplementary Figs. 1-3).

Baseline to 3 months post-intervention (n = 9): A total of nine participants completed the full structured training programme and the primary endpoint assessment (three months post-intervention), providing the data for the main statistical evaluation.

Six-month follow-up (n = 8): Eight participants remained available for the final longitudinal screening six months after the cessation of training. The reduction in sample size reflects the expected attrition rate typical of longitudinal studies involving neurodegenerative populations. All participants were encouraged to participate in social life and remain active after the intervention part of the programme. Six participants reported reading books, completing crosswords, involvement in a religious group, and gardening. One person reported taking a part-time job to stay active. Two participants did not initiate additional activities. These were the participants whose cognitive scores declined most significantly.

The results of the California Verbal Learning Test (CVLT) (Figs. 1, 2), the Attention and Perceptiveness Test (TUS), and the Hospital Anxiety and Depression Scale (HADS) were not entirely consistent across subjects. No statistically significant differences were found between test results.

The results of Addenbrooke’s Cognitive Examination Third Edition (ACE-III) for single domains were found to be significant in 4 out of 5 domains for pre- and post-intervention screening (attention: z = –2.100, p = 0.036; memory: z = –2.366, p = 0.021; fluency: z = –2.201, p = 0.034; visuo-spatial: z = –2.201, p = 0.031) (Table 2). The overall ACE-III score was also significantly higher across all follow-up assessments, including 6 months after the intervention (I-II scoring: z = –2.666, p = 0.009; I-III scoring: z = –2.547, p = 0.013; I-IV scoring: z = –2.547, p = 0.013) (Table 2). The effect size for all ACE-III comparisons was moderate (Table 2).

Discussion

The study analysed the results of a structured paper-pencil-based cognitive intervention and the potential prolonged effect of cognitive training among patients with late-onset AD. Participants’ cognitive scores on a screening test (ACE-III) improved immediately after the 12-week intervention and remained higher even after six months without intervention for six participants (Table 2).

Despite promising data from a pilot study (Segiet et al. 2025), scores on the CVLT and TUS did not show a consistent distribution and varied across participants. Due to the small number of participants, the moderation of depression, anxiety and CRIq score was not possible.

Despite the lack of significant results on detailed neuropsychological tests, there was also no significant deterioration, which, in the case of neurodegenerative diseases characterised by a gradual progression of symptoms, can be considered a very positive effect from a clinical point of view.

The observed statistical significance in global ACE-III scores, juxtaposed with the relative stability of domain-specific neuropsychological measures (CVLT, TUS, and HADS), suggests a preliminary signal of cognitive response. While more detailed cognitive screening tools remain sensitive to the neurodegenerative changes inherent in AD, the observed maintenance of global screening performance may reflect an initial, albeit modest, neuroplastic response to the intervention’s novelty and challenge (Belleville et al. 2011).

This suggests that pilot, structured cognitive training based on EL could serve as a catalyst for functional compensation, encouraging the use of broader neural networks to support overall performance (Grzenda et al. 2024; Livingston et al. 2020; Zając-Lamparska 2018). From a clinical perspective, these preliminary observations are encouraging. They suggest that at least time-limited stabilisation of functioning may be achievable even when specific deficits persist. Such a buffer against rapid decline could potentially support the preservation of independence and quality of life by facilitating compensatory strategies.

Crucially, the interpretation of the study’s long-term effects is significantly moderated by the participants’ post-intervention behaviours. Interview data revealed that seven of the 9 participants initiated or sustained cognitively stimulating activities – including reading, crosswords, and community engagement – following the primary training. Notably, the two participants who did not engage in such activities exhibited the most pronounced longitudinal cognitive decline.

While this suggests a potential synergistic effect between the paper-pencil training and increased self-efficacy, it introduces a critical methodological confounding factor that limits the isolation of the primary intervention’s prolonged efficacy. Given the small sample size (N = 9), the observed stability in global functioning among the active cohort may be more appropriately attributed to these self-initiated lifestyle modifications than to the residual effects of the cognitive training itself. Consequently, these findings should be interpreted as a signal that the intervention may serve as a behavioural primer for sustained mental activity, rather than as a stand-alone source of long-term cognitive improvement. Future research employing randomised controlled designs is necessary to establish the unique contribution of structured training on cognitive functioning.

Conclusions

These observations and positive feedback from participants underscore the need for activities in people’s lives. Therefore, professionals should encourage both individuals and their families to engage in daily chores and cognitive stimulation as a holistic approach to supporting people with cognitive decline (Newton et al. 2021). Engaging in new cognitive challenges may also encourage people with AD to remain socially active, giving them a sense of empowerment and helping them maintain cognitive abilities, such as carrying on a conversation or working around the house or garden with their families, as suggested by the study participants’ experiences of attending a religious group. It would also be in line with the LANCET Commission findings and help eliminate the threat to cognitive performance posed by isolation (Livingston et al. 2020).

Daily stimulation of cognitive activity can improve not only individuals’ quality of life but also the effectiveness of pharmacological treatment. It appears that a wide range of cognitive challenges and activities is sufficient to provide benefits even to people with severe cognitive impairment (Newton et al. 2021).

Limitations of the study

The main limitations of this study are its small sample size and quasi-experimental design. Small groups reduce the possibility of generalizability, increase the risk of statistical errors (type I and II), and make the results more sensitive to individual differences. The p-value may also be less reliable. Quasi-experimental designs are susceptible to confounding variables due to limited control over external factors. Despite efforts, fully isolating the long-term effect of cognitive training is challenging.

Recruiting participants was difficult because many declined due to time commitments, commuting, or post-pandemic concerns, despite assurances that participation was voluntary. The study included only patients with late-onset AD, which can be seen as both a limitation and a strength. While this excludes younger individuals, it provides a more homogeneous group with a more predictable cognitive profile, improving the consistency of observed effects. Furthermore, self-initiated activities reported by participants may represent a significant confounding factor. In a quasi-experimental design, it was impossible to isolate the prolonged effectiveness of the paper-pencil intervention from the neuroprotective benefits of the subsequent lifestyle changes. The observed stability in the 7 active participants may be more the result of “active maintenance” than a residual effect of the primary training alone. Further studies should take those concerns into account to determine the sole significance of the cognitive training.

Disclosures

A grant from the Priority Research Area under the Strategic Programme Excellence Initiative at Jagiellonian University supported the research presented in this publication.

The study was approved by the Bioethics Committee of the Jagiellonian University Medical College (Approval No. 1072.6120.306.2022).

The authors declare no conflict of interest.

Supplementary material is available on the journal’s website.

References