Psychological Medicine

Original Articles

Efficacy and specificity of computer-assisted cognitive remediation in schizophrenia: a meta-analytical study

O. Grynszpana1a2 c1, S. Perbala1, A. Pelissoloa1, P. Fossatia1a2, R. Jouventa1a2, S. Dubala1 and F. Perez-Diaza1

a1 Centre Emotion, CNRS USR 3246, Hôpital de La Salpêtrière, Paris, France

a2 Pierre et Marie Curie University, Paris, France


Background Cognitive remediation is frequently based on computerized training methods that target different cognitive deficits. The aim of this article was to assess the efficacy of computer-assisted cognitive remediation (CACR) in schizophrenia and to determine whether CACR enables selective treatment of specific cognitive domains.

Method A meta-analysis was performed on 16 randomized controlled trials evaluating CACR. The effect sizes of differences between CACR and control groups were computed and classified according to the cognitive domain assessed. The possible influences of four potential moderator variables were examined: participants' age, treatment duration, weekly frequency, and control condition type. To test the domain-specific effect, the intended goal of each study was determined and the effect sizes were sorted accordingly. The effect sizes of the cognitive domains explicitly targeted by the interventions were then compared with those that were not.

Results CACR enhanced general cognition with a mean effect size of 0.38 [confidence interval (CI) 0.20–0.55]. A significant medium effect size of 0.64 (CI 0.29–0.99) was found for Social Cognition. Improvements were also significant in Verbal Memory, Working Memory, Attention/Vigilance and Speed of Processing with small effect sizes. Cognitive domains that were specifically targeted by the interventions did not yield higher effects than those that were not.

Conclusions The results lend support to the efficacy of CACR with particular emphasis on Social Cognition. The difficulty in targeting specific domains suggests a ‘non-specific’ effect of CACR. These results are discussed in the light of the possible bias in remediation tasks due to computer interface design paradigms.

(Received March 20 2009)

(Revised October 10 2009)

(Accepted October 13 2009)

(Online publication April 12 2010)


c1 Address for correspondence: Dr O. Grynszpan, CNRS USR 3246, Service du Professeur Roland Jouvent, Pavillon Clérambault, Bâtiment de La Force, Hôpital de La Salpêtrière, 47 Bd de l'Hôpital, F-75013 Paris, France. (Email: