Research Article: Prediction during statistical learning, and implications for the implicit/explicit divide

Date Published: May 21, 2012

Publisher: University of Finance and Management in Warsaw

Author(s): Rick Dale, Nicholas D. Duran, J. Ryan Morehead.


Accounts of statistical learning, both implicit and explicit, often invoke
predictive processes as central to learning, yet practically all experiments
employ non-predictive measures during training. We argue that the common
theoretical assumption of anticipation and prediction needs clearer, more direct
evidence for it during learning. We offer a novel experimental context to
explore prediction, and report results from a simple sequential learning task
designed to promote predictive behaviors in participants as they responded to a
short sequence of simple stimulus events. Predictive tendencies in participants
were measured using their computer mouse, the trajectories of which served as a
means of tapping into predictive behavior while participants were exposed to
very short and simple sequences of events. A total of 143 participants were
randomly assigned to stimulus sequences along a continuum of regularity.
Analysis of computer-mouse trajectories revealed that (a) participants almost
always anticipate events in some manner, (b) participants exhibit two stable
patterns of behavior, either reacting to vs. predicting future events, (c) the
extent to which participants predict relates to performance on a recall test,
and (d) explicit reports of perceiving patterns in the brief sequence correlates
with extent of prediction. We end with a discussion of implicit and explicit
statistical learning and of the role prediction may play in both kinds of

Partial Text

To what extent is prediction related to sequential learning and memory, and to
implicit or explicit knowledge of that learning? In this paper, we offer a novel
methodology that may help answer this question, and present experimental results
that suggest this methodology holds promise for connecting these phenomena:
prediction, statistical learning, and explicit awareness. In brief, our experiment
is a simple manual spatial-position tracking task, in which a participant’s
behavior is tracked with the computer-mouse cursor. We are thus able to detect
predictive movements readily. We show that predictive behaviors emerge quickly in a
simple short-sequence design, using 48-element sequences of varying grammatical
regularity. Prediction, learning, and explicit knowledge all correlate strongly.

Here, we use an experimental design that reveals manual prediction, and we
investigate the properties of this behavior. To provide an unambiguous measure of
prediction, we turned to the measurement of hand movement during task performance.
In several recent studies, the semi-continuous movements of the computer-mouse
cursor were regarded as a direct and (occasionally) uninterrupted translation of
unfolding cognitive processes (Song & Nakayama,
2009; Spivey, Grosjean, & Knoblich,
2005). Motivated by this logic, we tracked participants’ computer
mouse as they clicked on a visual cue that moved around a spatial landscape on the
computer screen. Every time the cue was clicked, it momentarily disappeared before
reappearing in a new location. During this period of disappearance, the learner had
an opportunity to predict the most likely region of reappearance of the cue (it is
like a simplified version of the “Whac-A-Mole” classic arcade game
that readers may be familiar with). By tracking the coordinates of the
computer-mouse, results can show when (or if) participants manually gravitate
towards predictable regions. We used this in a simple sequence-learning task, in the
spirit of Nissen and Bullemer (1987) and of
Hunt and Aslin (2001). The task requires
participants to respond to spatial stimuli that occur in sequences that vary in
their ordering regularities.

Admittedly, we designed a very simple task, and used it to explore
initial response tendencies. Results thus reflect the
processing of short-term event sequences that may be routinely faced by cognitive
systems during daily activities (e.g., observing or producing brief structured
action sequences; Botvinick & Plaut,
2004). Our experiment simplified this ecological context, and exposed
participants to a single stream of visuospatial information. Certainly, the
experiment is not of the same scope of traditional statistical learning and of SRT
tasks, which use more complex sequences over many blocks of training. In that
respect, what we are revealing is the very beginning of the learning system’s
behavior, using computer-mouse trajectories to unveil the
“microstructure” of this initial processing.




0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments