Cognitive rehabilitation and cognitive training for mild dementia


Dr Masaru Mimura MD PhD, Department of Neuropsychiatry, Showa University School of Medicine, 1-5-8 Hatanodai, Shinagawa-ku, Tokyo 142-8666, Japan. Email:


Cognitive deficits caused by dementing illnesses are chronic and progressive problems, which should be tackled both by biological and non-biological approaches. Among the various techniques of non-biological approaches (cognitive rehabilitation), centered is cognitive training intervention for individuals with dementia. Cognitive training is further divided into two different types of setting: group and individualized. Among group training techniques, the reality orientation training and day care/day services are known to have evidence-based efficacy. Individually tailored cognitive training aims to directly and explicitly improve cognitive functioning of people with dementia specifically in the early stages. Increasing evidence demonstrates the efficacy of various individualized training programs for dementia, including Alzheimer's disease (AD). Specifically, three techniques, known as spaced retrieval, dual cognitive support and procedural memory training, have shown promise in their ability to enhance learning in people with dementia. In addition, recent studies have suggested that a combination of pharmacotherapy and cognitive training may benefit individuals with AD. Cognitive training appears to be particularly effective for people with mild memory impairment who are on cholinergic treatment. It is now widely accepted that the theoretical framework of ‘errorless learning’ is also a guiding principle in the realm of cognitive training for people with dementia. Although the effect of factors, namely effort (effortful vs effortless) and stimulus features (perceptual vs conceptual), has not been fully determined, error elimination during learning sessions is essential for favorable outcomes.


Rehabilitation implies the restoration of patients to the highest level of physical, psychological and social adaptation attainable. It includes all measures aimed at reducing the impact of disabling and handicapping conditions and at enabling disabled people to achieve optimum social integration.1 In addition, rehabilitation is not limited to therapeutic activities performed by patients themselves. Rather, it includes a process whereby people who are disabled by injury or disease work together with professional staff, relatives and members of the wider community to achieve their optimum physical, psychological, social and vocational well-being.2

Thus, rehabilitation for disabling and handicapping conditions consists of a wide range of interventions. Characteristics of such holistic and multifaceted approach of rehabilitation are even more prominent in the case of cognitive rehabilitation. Cognitive rehabilitation is an approach to help people with cognitive impairment work together with health care professionals to identify personally relevant goals and to devise strategies for addressing these issues.3 The emphasis is not on enhancing performance on cognitive tasks as such, but on improving functioning in the everyday context.4 The cognitive rehabilitation approach has been developed mainly through work with younger individuals with traumatic brain injury orstroke patients,5,6 but is now increasingly being discussed in relation to dementia.7,8 Increasing evidence validates cognitive rehabilitation programs for dementia, including Alzheimer's disease (AD).


Cognitive deficits caused by dementing illnesses are chronic and progressive problems, which should be tackled both by biological and non-biological approaches. By definition, the non-biological approaches (cognitive rehabilitation) include not only cognitive intervention for individuals with dementia (cognitive training), but also adjustment of patients' circumstances and a support system for their families and caregivers (Fig. 1). Cognitive training, an intervention to ameliorate cognitive deficits in patients with dementia, may further be divided into two different types of setting: group training and individualized training. In the group approach, each individual is assumed to have homogeneous and consistent impairments. Group training works on focused disabilities with the presumption that each individual similarly acts on intervention techniques. Procedures frequently used in a clinical setting as group cognitive training include reality orientation training, reminiscence therapy, music therapy, art therapy, validation therapy, day care/day services. Of these techniques, reality orientation training and day care/day services are known to have evidence-based efficacy.9,10 In contrast, evidence-based findings do not provide strong support for the use of reminiscence therapy, music therapy, art therapy and validation therapy, either due to methodological limitations or because of a lack of any randomized control trials in these areas.11–13

Figure 1.

Biological and non-biological therapeutic approaches for treating people with dementia.

Another approach of cognitive intervention is individually tailored cognitive training, which is planned to help people with dementia specifically in the early stages of the disease. Although any intervention strategy or technique may indirectly and eventually enhance cognitive performance of people with dementia (e.g. by relaxing a patient's tension), individualized cognitive training aims to directly and explicitly target cognitive functioning.


Observational studies have found that people who participate in cognitively stimulating leisure activities, such as reading, playing board games and playing musical instruments, have a lower risk of developing dementia.14 Accordingly, cognition-based approaches potentially work to prevent or slow the progression of the dementia process. Cognitive training involves practicing tasks in order to improve various elements of cognitive functioning. The underlying assumption is that practice may improve, or at least maintain, functioning in the given domain and that any effects of practice will generalize beyond the immediate training context.15 Cognitive training is designed for repeated practice of a set of standard tasks reflecting particular cognitive functions, such as memory, attention or problem-solving abilities. Among the wide range of cognitive domains, memory, and acquisition of novel information in particular, is most frequently the focus of cognitive training.16

Individualized cognitive training including memory training is offered through individual sessions.17–20 The training is occasionally facilitated by family members21,22 with therapist support, homework tasks being a popular method. Tasks may be presented in paper-and-pencil17,18,21,22 or computerized23–25 form, or may involve analogs of activities of daily living.19,26 Usually a range of difficulty levels is available within a standardized set of tasks to allow for selection of the level of difficulty that is most appropriate for a given individual.

Three techniques, known as spaced retrieval, dual cognitive support and procedural memory training, have shown promise in their ability to enhance learning in people with dementia. The spaced-retrieval (extended rehearsal) technique27,28 involves learning trials where a specific stimulus (e.g. a face) and a specific association (e.g. a name) are presented. Learning trials are separated by progressively longer time intervals filled with conversation or mental tracking tasks to prevent rehearsal of the to-be-remembered information. If an error occurs on retrieval (e.g. incorrect retrieval of a name when a face is presented), corrective feedback is provided and the interval between stimulus presentation and recall is decreased to the previous interval in which recall was correct. It has been postulated that the spaced-retrieval technique works by engaging implicit memory processes, by tapping procedural systems and by decreasing reliance on semantic or declarative mechanisms.29

Dual cognitive support involves the provision of cues and the enhancement of the saliency and organization of the to-be-remembered information at both encoding and retrieval of the information. Such support appears to be associated with improved learning in AD.30–32 Procedural memory training requires the activation of the motor system. In AD, motor learning has been shown in paradigms that require the self-selection of movements.33–35

Although specific cognitive function can be trained separately in mildly or moderately impaired dementia, an integrated treatment package by combining each technique of spaced retrieval, dual cognitive support and functional procedural skills training appears to be more powerful in enhancing cognitive activities of AD individuals, as was demonstrated by Loewenstein et al.36


Pharmacotherapy using cholinergic agents is now widely used in the possible prevention of the progression of AD. Recent studies have suggested that a combination of pharmacotherapy and cognitive training may benefit individuals with AD. In accordance with the suggestion that cognitive training may enhance the effects of antidementia drug therapy,37 some studies have evaluated the efficacy of cognitive training in combination with acetylcholinesterase inhibitors or other medications.18,38Figure 2 illustrates the efficacy of individual cognitive rehabilitation together with anticholinesterase medication.

Figure 2.

Comparison between the cognitive training (CT) group + acetylcholinesterase inhibitor (AChEI) compared with the AChEI alone. Performance on memory test after 3 months. SD, standard deviation; CI, confidence interval; ADAS-Cog, Alzheimer's disease assessment scale, cognition. Reproduced with permission from Clare et al.4

Cognitive training is particularly effective for people with mild memory impairment who are on cholinergic treatment. Loewenstein et al.36 evaluated the efficacy of a cognitive training program on memory and other cognitive functions of mildly impaired AD patients receiving a cholinesterase inhibitor. Cognitive training sessions included face–name association, object recall training, everyday functional tasks (e.g. making change, paying bills), orientation to time and place, visuomotor processing speed and the use of a memory notebook. Gains in recall of face–name associations, orientation, cognitive processing speed and specific functional tasks were present postintervention and at a 3 month follow up. A systematic program of cognitive training may enhance the pharmacological effect of maintaining performance on specific cognitive and functional tasks in mildly impaired AD patients.


Among the theoretical frameworks of memory training, ‘errorless learning’ is the most widely known concept. Errorless learning originally stems from the work of animal discrimination learning,39 in which pigeons were found to show better learning of red–green visual discrimination when prevented from making errors during training using a fading cue procedure than when trained using a traditional trial-and-error procedure. Several attempts have been made since then to apply this technique to teaching new skills to people with learning disabilities.40

More recently, a special interest has centered on the applicability of the technique to the rehabilitation of memory impaired individuals. Baddeley and Wilson41 demonstrated that amnesic patients, as well as control participants, learned a list of words under the errorless learning condition more effectively than the ‘errorful’ learning condition. The advantage of errorless learning was especially pronounced for the amnesic group. Wilson et al.42 further confirmed the greater benefit from errorless learning in a series of single case studies in which memory impaired patients with different etiologies learned a variety of materials, including names of objects and people, orientation items, items of general knowledge and how to program an electronic aid.

Several researchers have argued that errorless learning is not a specific technique, but a guiding principle.43,44 The interventions involved adaptation of learning methods for which there was prior evidence of potential usefulness.27,45–47 The results suggested that this approach may be beneficial for a proportion of people with early stage AD.

Based on their preliminary evidence for the effectiveness of cognitive rehabilitation interventions in early stage AD,48–50 Clare et al.51 further explored errorless learning principles in a controlled trial. Twelve participants meeting criteria for probable AD, with MMSE scores of 18 or above, were trained in face–name associations using an errorless learning paradigm. Under the principle of errorless learning, the training method involved selecting a mnemonic, learning the name using vanishing cues and rehearsing the name using spaced retrieval (expanding rehearsal). Training produced a significant group improvement in recall of trained, but not control, items (Fig. 3). Gains were largely maintained 6 months later, in the absence of practice. There were differences in individual responses to intervention.

Figure 3.

Efficacy of errorless learning in people with Alzheimer's disease. Mean free-recall scores for trained (lighter columns) and untrained (darker columns) items at initial and postintervention (Post) and at 1, 3, 6 and 12 months follow up (m fu). Six face-name pairs were selected for the training set and the remaining six pairs were allocated to the control set. For initial and postintervention scores of trained items, P < 0.001, whereas for initial and postintervention scores of untrained items, P > 0.05. Reproduced with permission from Clare et al.51


As pointed out by Riley and Heaton,52 traditional applications of errorless learning comprise two components, namely error elimination and cue fading. Yet, its applications to amnesia differ in the emphasis on these two components. Baddeley and Wilson41 emphasized the error elimination component. In the errorful condition in their experiments, participants were encouraged to guess a target word in response to its word stem, resulting in the occurrence of incorrect responses. In the errorless condition, the participants were immediately presented with a target word, giving no opportunity to make incorrect responses. The error elimination procedure adopted by Wilson et al. entails the intact presentation of a target item to ensure that neither guess nor error is made during training. For this reason, Riley et al.53 refer to this procedure as errorless learning without fading (ELWF). In contrast, cue fading refers to a procedure in which cues or prompts are gradually faded according to the progress of learning. This procedure has been introduced into the intervention of memory impaired individuals and referred to as the method of vanishing cues (MVC) by Glisky et al.54

Errorless learning without fading guarantees that no errors are made by providing a complete cue throughout a training period. Conversely, the intact presentation of a target results in boredom, thereby hindering the active participation of patients, which constitutes requisites for successful memory rehabilitation. Based on the generation effect,55 generating a target is assumed to require more effort and the more effort recruited, the better the memory performance.56 The MVC condition is expected to be most effective, because this condition is assumed to involve errorless and effortful processes, both of which contribute to superior performance.

In explicating such a trade-off, Komatsu et al.57 proposed a two-factor model in which the error (errorless vs errorful) and the effort (effortless vs effortful) factors are crossed. The results of their experiments with Korsakoff amnesics revealed that the ELWF and MVC conditions were more effective than the errorful learning conditions, thereby demonstrating an advantage of errorless learning. However, in constrast with the expectation, the effort factor was found to have little effect. More specifically, the number of errors under the MVC condition, although fewer than that under the errorful conditions, was still substantially above zero. Accordingly, the MVC condition should be viewed as a relatively, rather than absolutely, errorless task. It is likely that the ELWF and MVC conditions correspond to the distinction between error elimination and error reduction, respectively, and that learning is most effective when errors are completely eliminated than when relatively reduced.58

We subsequently introduced a revised version of the MVC procedure and further explored the effects of error and effort on learning by amnesic patients. We revised the MVC in such a way that the number of vanishing stages is increased using perceptually degraded stimuli as vanishing cues (Fig. 4). As was expected, our preliminary findings with the revised MVC, which emphasized gradual fading, demonstrated error elimination while maintaining effortful processes resulting in better performance.59

Figure 4.

Four study sessions crossing error (errorless vs errorful) and effort (effortless vs effortful) factors. MVC, method of vanishing cues; ELWF, errorless learning without fading. Reproduced with permission from Mimura et al.59


Recently, in examining the effectiveness of errorless learning, Tailby and Haslam60 developed yet another study condition where participants were given a verbal description of a target item and were asked to generate it. Such semantic generation was deemed as errorless in that the semantically rich description of a target word does not allow participants to generate words other than the target. Participants with memory impairment learned a list of words under each of the semantic generation, ELWF and errorful conditions, the latter two being comparable to those by Baddeley and Wilson.41 Memory performance under the two errorless conditions was better than that under the errorful condition, demonstrating an advantage of errorless learning. Most important, performance under the semantic generation condition was better than that under the ELWF condition. In terms of the effort factor, the semantic generation condition is placed as effortful, because the generation of a target word in response to its verbal description cue is thought to contribute to the elaborative processing of a target that is not presented in its intact form during training. Therefore, the semantic generation method is assumed to be comparable to the revised MVC. Notwithstanding the similarity, the two methods differ in stimulus features they rely on: perceptual features are critical in the revised MVC where participants are required to produce a target item prompted with perceptually degraded cues; in contrast, conceptual features, such as word meaning, are important in the semantic generation method where participants were asked to generate a target prompted by its verbal description. In our preliminary report, individuals with AD learned a list of words under four different study conditions that were designed by crossing the error (errorless vs errorful) and the feature (perceptual vs conceptual) factors. As was expected, individuals with AD learned best under the errorless and conceptual study condition.61


Various interventions to ameliorate cognitive impairment in mild dementia have been discussed. Recent studies of cognitive training have focused on individuals with mild dementia. Growing evidence suggests that cognitive training is effective against functional deterioration in individuals with mild dementia. However, the evidence is far from conclusive and we should await future studies concerning the effect of cognitive training for people with mild dementia.


This study was supported, in part, by a Showa University Grant-in-Aid for Innovative Collaborative Research Projects.