Temporal Predictive Codes for Spoken Words in Auditory Cortex

Humans can recognize spoken words with unmatched speed and accuracy. Hearing the initial portion of a word such as “formu…” is sufficient for the brain to identify “formula” from the thousands of other words that partially match 1Marslen-Wilson W. Tyler L.K. The temporal structure of spoken language understanding.Cognition. 1980; 8: 1-71Crossref PubMed Scopus (840) Google Scholar, 2Zwitserlood P. The locus of the effects of sentential-semantic context in spoken-word processing.Cognition. 1989; 32: 25-64Crossref PubMed Scopus (332) Google Scholar, 3O'Rourke T.B. Holcomb P.J. Electrophysiological evidence for the efficiency of spoken word processing.Biol. Psychol. 2002; 60: 121-150Crossref PubMed Scopus (71) Google Scholar, 4Pulvermüller F. Shtyrov Y. Ilmoniemi R.J. Marslen-Wilson W.D. Tracking speech comprehension in space and time.Neuroimage. 2006; 31: 1297-1305Crossref PubMed Scopus (68) Google Scholar, 5Revill K.P. Aslin R.N. Tanenhaus M.K. Bavelier D. Neural correlates of partial lexical activation.Proc. Natl. Acad. Sci. USA. 2008; 105: 13111-13115Crossref PubMed Scopus (46) Google Scholar, 6Righi G. Blumstein S.E. Mertus J. Worden M.S. Neural systems underlying lexical competition: an eye tracking and fMRI study.J. Cogn. Neurosci. 2010; 22: 213-224Crossref PubMed Scopus (56) Google Scholar. Two alternative computational accounts propose that partially matching words (1) inhibit each other until a single word is selected (“formula” inhibits “formal” by lexical competition 7McClelland J.L. Elman J.L. The TRACE model of speech perception.Cognit. Psychol. 1986; 18: 1-86Crossref PubMed Scopus (1948) Google Scholar, 8Norris D. McQueen J.M. Shortlist B: a Bayesian model of continuous speech recognition.Psychol. Rev. 2008; 115: 357-395Crossref PubMed Scopus (408) Google Scholar, 9Gaskell M.G. Marslen-Wilson W.D. Integrating form and meaning: a distributed model of speech perception.Lang. Cogn. Process. 1997; 12: 613-656Crossref Scopus (379) Google Scholar) or (2) are used to predict upcoming speech sounds more accurately (segment prediction error is minimal after sequences like “formu…” 10Harris Z. From phoneme to morpheme.Language. 1955; 31: 190-222Crossref Google Scholar, 11Cairns P. Shillcock R. Chater N. Levy J. Bootstrapping word boundaries: a bottom-up Corpus-based approach to speech segmentation.Cogn. Psych. 1997; 33: 111-153Crossref PubMed Scopus (100) Google Scholar, 12Elman J.L. Finding Structure in time.Cogn. Sci. 1990; 14: 179-211Crossref Google Scholar). To distinguish these theories we taught participants novel words (e.g., “formubo”) that sound like existing words (“formula”) on two successive days 13Davis M.H. Di Betta A.M. Macdonald M.J.E. Gaskell M.G. Learning and consolidation of novel spoken words.J. Cogn. Neurosci. 2009; 21: 803-820Crossref PubMed Scopus (207) Google Scholar, 14Dumay N. Gaskell M.G. Sleep-associated changes in the mental representation of spoken words.Psychol. Sci. 2007; 18: 35-39Crossref PubMed Scopus (329) Google Scholar, 15Gaskell M.G. Dumay N. Lexical competition and the acquisition of novel words.Cognition. 2003; 89: 105-132Crossref PubMed Scopus (284) Google Scholar, 16Tamminen J. Payne J.D. Stickgold R. Wamsley E.J. Gaskell M.G. Sleep spindle activity is associated with the integration of new memories and existing knowledge.J. Neurosci. 2010; 30: 14356-14360Crossref PubMed Scopus (319) Google Scholar. Computational simulations show that knowing “formubo” increases lexical competition when hearing “formu…”, but reduces segment prediction error. Conversely, when the sounds in “formula” and “formubo” diverge, the reverse is observed. The time course of magnetoencephalographic brain responses in the superior temporal gyrus (STG) is uniquely consistent with a segment prediction account. We propose a predictive coding model of spoken word recognition in which STG neurons represent the difference between predicted and heard speech sounds. This prediction error signal explains the efficiency of human word recognition and simulates neural responses in auditory regions.