While the effects of practice on problem solving are easily observed using behavioral measures, attempts to build and validate models explaining this speedup have been limited by difficulty distinguishing quickly executed cognitive processes (e.g., Anderson, 1982; Logan, 1988; Rickard, 1997). In this experiment, we explore the three-phase model of skill acquisition that describes a participant's problem-solving strategy transition from computation to retrieval, and, with practice, to an automatic stimulus-response process (Fitts & Posner, 1967; Tenison et al., 2016). We collected magnetoencephalography (MEG) data while participants practiced solving a set of repeated math problems. We hypothesized that the processes of encoding, familiarity, and recollection are early features that distinguish the three phases of skill acquisition. To test this hypothesis we used two sets of analyses, one to identify the timing of these distinctions and one to test brain activation in meaningful locations in the cortex. Using classification techniques, we identify three time points during the first 700 ms of problem solving when there is notable improvements in our ability to distinguish the Learning phases. Finally, our analysis of 8 brain regions provided evidence that the three learning phases are distinguished both by the cognitive processes used in encoding and the learning mechanisms supporting within-phase learning

Tenison, C., Fincham, J., & Anderson, J. (In prep). From computation to automization: How practice alters initial neural response to familiar arithmetic problems.

  • Magnetoencephalography (MEG)
  • Analysis of both sensor and source space
  • Ridge regression classification with GLMNET
  • Mixed-effects modeling
  • Dr. John R. Anderson
  • Dr. Jon Fincham