Progress in Neurobiology 79 (2006) 49–71 www.elsevier.com/locate/pneurobio Language outside the focus of attention: The mismatch negativity as a tool for studying higher cognitive processes Friedemann Pulvermu¨ller *, Yury Shtyrov Medical Research Council, Cognition and Brain Sciences Unit, 15 Chaucer Road, Cambridge CB2 2EF, UK Received 20 June 2005; received in revised form 18 November 2005; accepted 4 April 2006 Abstract Which aspects of language and cognitive processing take place irrespective of whether subjects focus their attention on incoming stimuli and are, in this sense, automatic? The Mismatch Negativity (MMN), a neurophysiological brain response recorded in the EEG and MEG, is elicited by attended and unattended stimuli alike. Recent studies investigating the cognitive processes underlying spoken language processing found that even under attentional withdrawal, MMN size and topography reflect the activation of memory traces for language elements in the human brain. Familiar sounds of one’s native language elicit a larger MMN than unfamiliar sounds, and at the level of meaningful language units, words elicit a larger MMN than meaningless pseudowords. This suggests that the MMN reflects the activation of memory networks for language sounds and spoken words. Unattended word stimuli elicit an activation sequence starting in superior-temporal cortex and rapidly progressing to left-inferior-frontal lobe. The spatio-temporal patterns of cortical activation depend on lexical and semantic properties of word stems and affixes, thus indicating that the MMN can give clues about lexico-semantic information processing stored in long term memory. At the syntactic level, MMN size was found to reflect whether a word string conforms to abstract grammatical rules. This growing body of results suggests that lexical, semantic and syntactic information can be processed by the central nervous system outside the focus of attention in a largely automatic manner. Analysis of spatio- temporal patterns of generator activations underlying the MMN to speech may be an important tool for investigating the brain dynamics of spoken language processing and the activated distributed cortical circuits acting at long-term memory traces. # 2006 Elsevier Ltd. All rights reserved. Keywords: Lexicality; Syntax; Semantics; Cognitive neuroscience: Attention; ERP; MEG; EEG; Spatio-temporal patterns; Mismatch Negativity Contents 1. The mismatch negativity: Change detection and memory trace indicator . ....................................... 50 2. Motivation for investigating brain processes of language using the MMN. ....................................... 51 2.1. Automaticity ............................................................................. 51 2.2. Earliness . ............................................................................. 52 2.3. Suitability for studying individual linguistic items . ................................................ 52 2.4. Control of the effect of physical stimulus features: Identity MMN and theory . ............................... 54 3. Theoretical issues: Functional seriality or parallelism in spoken language processing? ............................... 54 3.1. Serial models of language processing ............................................................ 54 3.2. Neurophysiological evidence for serial information access in language comprehension . ....................... 55 3.3. Neurophysiological evidence for early psycholinguistic information access . ............................... 56 4. MMN evidence . ............................................................................. 56 4.1. Lexical access and selection . ................................................................ 57 4.2. Semantic processes ......................................................................... 60 4.3. Syntactic processes ......................................................................... 62 4.4. Varying attentional withdrawal in the study of language ............................................... 63 * Corresponding author. Tel.: +44 1223 355294x770; fax: +44 1223 359062. E-mail address: [email protected] (F. Pulvermu¨ller). 0301-0082/$ – see front matter # 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.pneurobio.2006.04.004 50 F. Pulvermu¨ller, Y. Shtyrov / Progress in Neurobiology 79 (2006) 49–71 5. Conclusions................................................................................... 64 5.1. The time course of information access in spoken language comprehension . ................................. 64 5.2. Change detection MMN and representational negativity . ............................................. 65 5.3. The contribution of fronto-central cortex to attention switching and language comprehension . .................... 66 6. Summary . ................................................................................... 67 Acknowledgement . ........................................................................... 67 References ................................................................................... 67 1. The mismatch negativity: Change detection and MMN. These have been documented by fMRI, EEG and MEG memory trace indicator studies (Opitz et al., 2002; Pulvermu¨ller et al., 2003; Rinne et al., 2000). Differences may exist between the generator Investigations of neuronal mass activity using electro- and constellations inferred from EEG and MEG recordings, magnetoencephalography (EEG, MEG) offer a unique oppor- especially if some of the local generators are radial. tunity to monitor non-invasively the spatio-temporal char- A main reason for the great importance of the MMN in acteristics of brain processes. One of the most frequently cognitive neuroscience is the insightthatitislinkedtohigher investigated neuronal mass responses in Cognitive Neu- order perceptual processes underlying stimulus discrimina- roscience is the Mismatch Negativity or MMN. The MMN tion, rather than being triggered simply by physical differences is an event-related response elicited by infrequent acoustic between standard and deviant stimuli (Na¨a¨ta¨nen et al., 1993; events, the so-called deviant stimuli, occasionally occurring Tervaniemi et al., 2001). Subjects exposed to a physical among frequently repeated sounds, called the standard stimuli. stimulus difference they are unable to discriminate may show Since the MMN was discovered (Na¨a¨ta¨nen et al., 1978), its only a minimal MMN, or even no MMN at all, to this acoustic properties were extensively studied using elementary acoustic contrast. However, as soon as the same subjects have learned to events, such as brief tone pips, with the rare deviant stimuli discriminate the critical stimuli, their MMN is reliably differing from the repetitive standard stimuli in pitch, length, triggered by the acoustic stimulus contrast and MMN size is loudness, etc. In these experiments, the MMN was found to be much increased (Kraus et al., 1995; Kujala et al., 2003; an indicator of acoustic change detection, which, surprisingly, Na¨a¨ta¨nen et al., 1993; Schro¨ger, 1996; Schro¨ger et al., 1994). was present regardless of whether subjects focussed their The MMN can even demonstrate a perception that does not attention on the stimuli or not. The MMN is therefore usually have a direct physical basis in the acoustic input. Physiological considered to be independent of focussed attention and it is, in manifestations of such perceptual illusions have been this sense, an automatic brain response (Na¨a¨ta¨nen et al., 2001; documented in a range of MMN studies (Colin et al., 2004; Na¨a¨ta¨nen and Winkler, 1999). When claiming that the MMN Micheyl et al., 2003; Ross et al., 1996; Stekelenburg et al., indicates ‘‘automatic change detection’’, it must, however, be 2004; Tervaniemi et al., 1994). For example, a tone with a gap taken into consideration that attentional load may influence the masked by broad-band noise elicits the so-called ‘‘continuity MMN (see Section 4.4) and that, in addition to its well-known illusion’’, the contra-factual perception of a continuous tone role as an acoustic change detection indicator, the MMN reveals bridging the actual gap. The amplitude of the MMN has been a range of higher cognitive processes. shown to reflect this illusory-continuity perception. Further- Similar to most neurophysiological mass phenomena, the more, the MMN could be demonstrated to reflect abstract MMN can be recorded in the electric and in the magnetic brain features of tone sequences. For example, in the context of response. In the latter case, it is also called the magnetic stepwise upwards moving standard tones, an occasional Mismatch Negativity or MMNm. Excitatory post-synaptic downwards step presented as the rare deviant stimulus elicits potentials arriving at the long apical dendrites of cortical an MMN irrespective of varying physical features of the pyramidal neurons give rise to an electric field that makes the sounds (Korzyukov et al., 2003; Paavilainen et al., 1995, 1998, cortex relatively negative at its surface and relatively positive at 2001; Saarinen et al., 1992; Tervaniemi et al., 1994). This and its basis (Mitzdorf, 1985). Concordant fields in thousands of similar evidence has led to the conclusion that, in addition to neurons can be picked up in the EEG signal and in the averaged changes in the acoustic environment, higher level cognitive event-related potential or ERP. The current flow that leads to processes in the auditory system are reflected