Success and failure of new speech category learning in adulthood: Consequences of learned Hebbian attractors in topographic maps

Abstract

One of the key issues that a theory of perception must address is the effect of prior experience. Newborn infants initially have the ability to distinguish a rich variety of linguistic contrasts, but by about six months of age, their ability has begun to attune to their native language (Kuhl, Williams, Lacerda, Stevens, & Lindblom, 1992). This reshaping of the perceptual space presumably allows the developing infants to recognize their language sounds more effectively and may contribute to the warping of perceptual representations toward category prototypes (the perceptual magnet effect; Kuhl, 1991Kuhl, , 2000 and to the sharpening of sensitivity at category boundaries (the categorical perception effect; Liberman, Harris, Hoffman, & Griffith, 1957). However, there is the suggestion that such reshaping may have the side effect of hindering the acquisition of language distinctions later in life (Flege, 1995). In this paper, we explore a mechanistic account of how first-language speech acquisition can influence the later acquisition of second-language speech sounds.Our exploration is guided by the assumption that speech learning is an instance of more general architectural and computational principles. We consider four such principles and evaluate their usefulness in addressing data from an experiment on the acquisition of a nonnative speech contrast in adulthood. Specifically, we use the principles to develop a computational model of the pattern of successes and failures in learning the American English / / and / / by adult native speakers of Japanese (McCandliss, Fiez, Protopapas, Conway, & McClelland, 2002). Models of perceptual learning often attempt to reproduce only the end state of learning, but frequently, this does not place sufficient constraints on the models (Damper & Harnad, 2000;Edelman & Intrator, 2002). In order to impose additional constraints, we evaluated our model against several measures over the time course of learning and under different training conditions and explored the extent to which the principles can account for a wide range of findings.The article is organized as follows. First, we give an overview of the R/L problem and of the McCandliss et al. (2002) results, pointing out aspects of the results that appear puzzling in the absence of an explicit mechanistic explanation. Next, we introduce and justify the principles underlying our model and present an abstract and simplified version to illustrate its basic properties. We then set out our model of the McCandliss et al. data and present the modeling results together with the experimental results. Finally, we evaluate the successes and limitations of the model and the consequent implications for the architectural principles. The R/L ProblemThe Japanese language does not have distinct / / and / / phonemes, and native speakers of Japanese have extreme difficulty distinguishing between the / / and / / sounds in American English (AE), particularly when they occur in syllable-initial positions (Logan, Lively, & Pisoni, 1991). When presented ...

Face-selective neurons maintain consistent visual responses across months

“…Development of perceptual systems in the brain is generally understood to involve a plastic phase that tunes the sensitivities of a naive network to match the statistics of sensory experience, followed by a stable phase in which functional properties are held fixed by local abatement of plasticity at synapses or through the global influence of attractor dynamics (36,37). Some evidence from longitudinal recordings from primary visual and somatosensory cortices supports this idea (26,38).…”

Phonetic learning as a pathway to language: new data and native language magnet theory expanded (NLM-e)

“…When processing non-native speech sounds, the adult brain is activated over a significantly longer duration and a significantly larger area than when processing native language sounds, showing that the processing of non-native sounds is neurally time consuming and requires additional brain resources (Zhang et al 2005). MEG studies also indicate that training adults on second language contrasts can be successful and is enhanced by the exaggeration of phonetic cues in a manner similar to motherese (Pisoni & Lively 1995;Iverson et al 2005;Vallabha & McClelland 2007;Zhang et al submitted).…”

The interaction between competition, learning, and habituation dynamics in speech perception

“…The unexpected result of the initial presentation order bias might also shed light on some results by Tuller, Jantzen, and Jirsa (2008), as well as by Vallabha and McClelland (2007). These researchers explicitly looked at the dependence of learning on the initial configuration of the perceptual space at the beginning of the experiment.…”

“…The perspective of dynamical systems has provided novel insights into numerous domains of human behavior, including movement coordination (e.g., Hock, Schöner, and Giese 2003), interpersonal coordination (e.g., Schmidt, Bienvenu, Fitzpatrick, and Amazeen 1998), learning (e.g., Zanone and Kelso 1992;Vallabha and McClelland 2007;Tuller, Jantzen, and Jirsa 2008), visual perception (e.g., Kawamoto and Anderson 1985), and speech perception (e.g., Grossberg and Myers 2000;Grossberg and Kazerounian 2011). With dynamical systems theory comes a commitment to look at the continuous aspects of perceptual and cognitive phenomena rather than purely looking at stable categories and discrete changes (Spivey 2007).…”

“…Thus, in the model that we are going to present, the processes underlying fast perceptual competition, slow learning, and slow habituation constrain each other in real time. We chose to investigate the interplay between these different processes because they provide explanations for the flexibility and plasticity of perception when studied in isolation (Kawamoto and Anderson 1985;Bogacz et al 2006;Vallabha and McClelland 2007). When these mechanisms are taken together, they not only predict response times during identification tasks, but also other, more long-term effects: as the model learns, its sensitivity to small acoustic differences is heightened, mirroring the way in which listeners' perception of ambiguous, unfamiliar, or degraded speech sounds becomes gradually more reliable with experience (Maye, Aslin, and Tanenhaus 2008;Bradlow and Bent 2008;Winters and Pisoni 2008;Samuel and Kraljic 2009).…”