Stimulus expectancy modulates inferior frontal gyrus and premotor cortex activity in auditory perception

Abstract

In studies on auditory speech perception, participants are often asked to perform active tasks, e.g. decide whether the perceived sound is a speech sound or not. However, information about the stimulus, inherent in such tasks, may induce expectations that cause altered activations not only in the auditory cortex, but also in frontal areas such as inferior frontal gyrus (IFG) and motor cortices, even in the absence of an explicit task. To investigate this, we applied spectral mixes of a flute sound and either vowels or specific music instrument sounds (e.g. trumpet) in an fMRI study, in combination with three different instructions. The instructions either revealed no information about stimulus features, or explicit information about either the music instrument or the vowel features. The results demonstrated that, besides an involvement of posterior temporal areas, stimulus expectancy modulated in particular a network comprising IFG and premotor cortices during this passive listening task.

Introduction

Auditory perception is associated with the primary and secondary auditory cortices, and depending on the spectral complexity of the perceived sounds, adjacent areas in the middle and posterior part of the superior temporal gyrus (STG). In addition to these areas, the perception of speech sounds is associated with extended temporal lobe activations, especially in the left hemisphere (Binder et al., 2000, Liebenthal et al., 2005, Meyer et al., 2005, Specht and Reul, 2003). However, frontal areas such as posterior parts of inferior frontal gyrus (pIFG), pars opercularis, and ventral premotor cortex (PMC) are also reported in some speech perception studies (Pulvermuller et al., 2006, Wilson et al., 2004), but the functional role of these areas is still debated. While Iacoboni (2008) advocates an essential role, Hickok (2009) argues that such areas have a modulatory rather than an essential role in speech perception. In a recent review, Price (2010) suggests an involvement of motor areas in speech perception primarily during the perception of degraded (e.g. noise perturbed) or sparse speech stimuli. Hence, one indication of such a compensatory role is that imaging studies are repeatedly revealing IFG and premotor activity during either the perception of noise perturbed stimuli (Osnes et al., 2011, Pulvermuller et al., 2006, Wilson et al., 2004), phonetic sounds that are foreign to the participants (e.g. Wilson & Iacoboni, 2006), or with sparse stimuli such as sine-wave replicas (SWS) of consonant-syllables (Dehaene-Lambertz et al., 2005, Dufor et al., 2007).

Ambiguous stimulus such as SWS provide the unique possibility to present the same physical stimulus as speech and non-speech, depending on the instructions and tasks given to the participants, e.g. discrimination tasks such as discriminating between specific consonant–vowel syllables and non-verbal sounds in two separate sessions (Dufor et al., 2007). With such a SWS paradigm, Dufor et al. (2007) report a more extensive left-lateralized network that involves lateral inferior frontal and parietal regions in addition to superior temporal regions after the participants were debriefed and thus perceived the same stimuli as speech and not longer as non-speech sounds. However, Giraud and Price (2001) as well as Indefrey and Cutler (2004) propose that active tasks involve additional cognitive processes such as working memory and attention, which are, besides of parietal areas, in particular associated with prefrontal areas, including IFG. Hence, the experimental setup may introduce a task confound that makes inferences about the functional role of frontal areas more difficult. In fact, IFG and ventral PMC have also been reported in relation to other processes, such as musical perception or action observation (Fadiga, Craighero, & D’Ausilio, 2009). Davis and Johnsrude (2007) argue that such task-driven processes interact with data-driven processes through both echoic and motoric/somatotopic representations of the speech sounds, in particular when processing unfamiliar or mismatching phonetic-category boundaries.

However, also situations where no overt task is performed the perceptual process can be influenced through top-down modulations, e.g. when passively listening to otherwise ambiguous sounds with explicit expectations of phonetic elements in the sounds. Thus, the expectation and the knowledge about the stimulus content may also modulate IFG and ventral PMC in addition to speech sound sensitive areas such as middle parts of the temporal lobe. Therefore, the aim of the present study was to investigate whether induced expectations of specific vowel or music-instrument features in spectral-mixed sounds would modulate the activity in these frontal and temporal areas. The stimuli were spectral mixes of either a flute and a vowel, e.g. /a/ (Vowel and flute mixes; VFM) or of a flute and a second instrument sound, e.g. trumpet (Music Instrument Mixes; MIM), presented in three consecutive fMRI sessions. The sound mixes were created so that the specific vowel or instrument sounds features, e.g. the trumpet or vowel /a/, were not easily recognizable when the actual stimulus creation procedure was unknown for the participant. Therefore, in the first session all participants were naïve and not aware that there were specific vowel or music features presented in the stimuli. In order to keep their attention constant throughout the experiment but without drawing their attention directly to the stimuli, the participants performed a task that was unrelated to any stimulus features (‘neutral instruction’), but only those trials, where no response was required, were used for the subsequent analyses. Prior to the second and third session stimulus expectations where induced by informing the participants about the fact that the stimuli were spectral mixes and that some of the sounds contained specific music sounds, like a trumpet sound (‘music instruction’) or a specific vowel sound, like the vowel /a/ (‘vowel instruction’), respectively. The participants listened also to an unmixed version of respective template sounds, i.e. a trumpet sound or the vowel /a/. The order of these two instructed sessions were cross-balanced across participants, and the participants performed no explicit task during these two sessions, other than passively listening to the sounds. We expected that debriefing the participants, and thus inducing expectations about the stimuli, will modulate the activation of frontal areas, in particular IFG and premotor cortex, in addition to auditory areas such as STG and planum temporale (PT).