The neural correlates of statistical learning in a word segmentation task: An fMRI study

Abstract

Functional magnetic resonance imaging (fMRI) was used to assess neural activation as participants learned to segment continuous streams of speech containing syllable sequences varying in their transitional probabilities. Speech streams were presented in four runs, each followed by a behavioral test to measure the extent of learning over time. Behavioral performance indicated that participants could discriminate statistically coherent sequences (words) from less coherent sequences (partwords). Individual rates of learning, defined as the difference in ratings for words and partwords, were used as predictors of neural activation to ask which brain areas showed activity associated with these measures. Results showed significant activity in the pars opercularis and pars triangularis regions of the left inferior frontal gyrus (LIFG). The relationship between these findings and prior work on the neural basis of statistical learning is discussed, and parallels to the frontal/subcortical network involved in other forms of implicit sequence learning are considered.

Keywords: Artificial language; Broca’s area; LIFG; Sequence learning; Statistical learning; Word segmentation; fMRI.

Figure 1

Exposure phase presented in a blocked design. In each of 4 runs, participants listened to 2 minutes each of forward, backward and overlaid speech streams. Each type of speech stream was paired with a unique color cue. 30-second blocks of language exposure were interleaved with 15-second periods of silence. The total duration of each run was approximately 8 minutes and 30 seconds.

Figure 2

Differences in word and partword rating for test items presented as forward speech and as backward speech. (A) Forward words were rated as significantly more familiar than forward partwords on each test. (B) Backward words were rated significantly higher than backward partwords only on test 3. (C) The rating difference between words and partwords presented in forward speech was significantly higher than the rating difference for items presented in backward speech.

Figure 3

Variability in learning. On each test, participants’ delta-learning scores across tests spanned a wide range (measured as change in forward W-PW rating for each test as compared to the test preceding it). The mean change in rating for each test is indicated in red.

Figure 4

Results of whole brain analysis with sagittal sections removed (A) Mean activation in response to forward > backward continuous speech streams (Z>2.3, p<0.05) (B) Neural changes related specifically to learning in the left inferior frontal gyrus as revealed by a delta analysis performed on the contrast forward >backward (Z>1.9, p<0.05).