Generative time telling in adults with disabilities: A matrix training approach

Curiel et al. (2020) worked with two adults who had developmental delays. The team wanted to see if matrix training could teach the adults to read any clock time without drilling every minute.

They picked 12 clock faces to teach directly. After the lessons they tested 132 new times the adults had never seen.

Both adults got about half of the new times right. Teaching 12 clock faces let them correctly name around 50 % of 132 untrained times.

The result shows partial generative learning. Some untaught times popped out without extra training.

Clements et al. (2021) got even bigger generative jumps. They taught 12 three-digit numbers and saw 8–12 new numeral tacts appear for each one taught. The clock study produced fewer untaught responses, but both used the same matrix-training logic.

Frampton et al. (2023) wrote a how-to guide that covers the 12×12 clock-time design used here. Their tutorial gives you the decision steps to copy or tweak this exact setup.

Lancioni et al. (2009) and Meier et al. (2012) also saw untaught tacts emerge, but they used mand training instead of matrix training. The takeaway: multiple roads can lead to new tacts; matrix training is simply one clear roadmap.

You can save hours by letting the matrix do the teaching. Pick a small set of clock faces that span the hour and minute hands, teach them with errorless prompts, then probe every five-minute combination. If your learner hits 50 % or better, you just cut your teaching set by 90 %. If not, add a few more targets and probe again. Either way, you spend less session time on rote drills and more on functional time-telling in daily routines.