The study took place in three middle schools in one school district. The interventions were administered in the students’ homes and the academic assessments were completed at school.
A total of 101 students in grades 6–8 were included in the study. Students were assigned to one of three study groups: a two-component working memory training (WMT; 33 students), an adaptive one-component WMT (33 students), and a non-adaptive one-component WMT (35 students). Approximately 44% of the students were male, 8% were receiving special education services, 6% had an Individualized Education Plan, and 5% had a learning disorder. Approximately 75% were White, 14% were Asian, 2% were Black, and 9% other race. Two percent were Hispanic or Latino.
Students in the intervention group used a version of the Cogmed WMT computer program called two-component WMT that was intended to improve long-term (or secondary) memory. The intervention was administered to students individually in their homes over the internet. In each training session, the student completed eight working memory exercises that asked students to recall a certain number of items (span length). Four exercises were “critical exercises” that students completed in every training session. Four exercises were a subset of eight “common exercises” that rotated across training sessions. Each of the exercises involved 15 trials with adaptive span length, meaning that the number of items the student was asked to recall in a trial increased as the student recalled items accurately. The common exercises had a 100% recall accuracy threshold (RAT)—that is, the student was first asked to recall two items; once they recalled two items with no errors, they were asked to recall three items; once they recalled three items with no errors, they were asked to recall four items; and so on. Researchers expected students to use their short-term (or primary) memory in completing these 100% RAT exercises. The critical exercises had a reduced accuracy threshold after three items—that is, 100% recall accuracy was required for two- and three-item spans, but the student was allowed to make one error on four- , five-, and six-item spans (for example, once the student recalled four items with no more than one error, they were asked to recall five items); two errors on seven-, eight-, and nine-item spans; three errors on 10-, 11-, and 12-item spans; and so on. Because the reduced RAT could enable a student to attain longer spans than they might attain with 100% RAT, the researchers expected students to use their long-term memory in completing the reduced-RAT exercises. Training sessions were intended to last 30–45 minutes each day, five school days per week, for 5 weeks. Students trained for 25 days, on average, and spent at least 75% of the last 10 days of training attempting to recall spans of four or more items in the critical exercises.
Students in the comparison groups used versions of the Cogmed WMT computer program that researchers expected to exercise their short-term memory but not their long-term memory. The comparison interventions were administered to students individually in their homes over the internet. As in the intervention group, the student completed eight exercises per training session, consisting of four critical exercises that did not change and four common exercises that rotated across sessions. In the adaptive one-component WMT comparison group, all eight exercises had an adaptive span length with 100% RAT, and each exercise consisted of 15 trials. In the non-adaptive one-component WMT comparison group, all eight exercises had a fixed span length of two items with 100% RAT, meaning students were always asked to recall two items with no errors; each exercise consisted of 20 trials. Students in the adaptive one-component WMT group trained for 25 days, on average, and spent at least 75% of the last 10 days of training attempting to recall spans of four or more items in the critical exercises. Students in the non-adaptive one-comparison WMT group trained for 25 days, on average.
Support for implementation
At the beginning of the study, all students in the three groups and their families were coached on how to use the WMT computer program. As students went through the program, they received weekly, scripted calls from trained coaches to check compliance with the program and to assist the student with any technical issues.