MUC (Memory, Unification, Control) and beyond.

Hagoort P. · 2013

Read MUC (Memory, Unification, Control) and beyond. by Hagoort P. (2013) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3709422.txt.

Introduction

Hagoort presents the MUC model of how the brain handles language, offered as a successor to the classical Wernicke-Lichtheim-Geschwind picture. MUC divides language processing into three components. Memory stores linguistic knowledge. Unification assembles words into structures and meanings. Control manages language use in context, such as choosing between languages.

Each component is tied to its supporting brain networks. Hagoort then argues the model must reach beyond the classic language regions. Attention systems and networks for inference are needed to explain how we grasp real discourse, beyond single words and literal content. The paper sets out that extended architecture.

Catalog record

CREATORHagoort P.
YEAR2013
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC3709422

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

MUC (Memory, Unification, Control) and beyond By Peter Hagoort Published in: Frontiers in Psychology DOI: 10.3389/fpsyg.2013.00416 ABSTRACT Abstract A neurobiological model of language is discussed that overcomes the shortcomings of the classical Wernicke-Lichtheim-Geschwind model. It is based on a subdivision of language processing into three components: Memory, Unification, and Control. The functional components as well as the neurobiological underpinnings of the model are discussed. In addition, the need for extension of the model beyond the classical core regions for language is shown. The attention network and the network for inferential processing are crucial to realize language comprehension beyond single word processing and beyond decoding propositional content. It is shown that this requires the dynamic interaction between multiple brain regions. Keywords: neurobiology of language, Memory, Unification, Control, speaker meaning, language connectivity INTRODUCTION An adequate neurobiological model of our uniquely human language faculty has to meet the following two requirements: (1) it should decompose language skills such as speaking and listening into the contributing types of knowledge and processing steps (the cognitive architecture); (2) it should specify how these are instantiated in, and supported by the organization of the human brain (the neural architecture). Until not too long ago, the neurobiological model that has dominated the field was the Wernicke-Lichtheim-Geschwind (WLG) model (see Figure 1). In this model, the human language faculty was situated in the left perisylvian cortex, with a strict division of labor between the frontal and temporal regions. Wernicke's area in left temporal cortex was assumed to subserve the comprehension of speech, whereas Broca's area in left inferior frontal cortex was claimed to subserve language production. The arcuate fasciculus connected these two areas. [Figure: The classical Wernicke-Lichtheim-Geschwind model of the neurobiology of language. In this model Broca's area is crucial for language production, Wernicke's area subserves language comprehension, and the necessary information exchange between these areas (such as in reading aloud) is done via the arcuate fasciculus, a major fiber bundle connecting the language areas in temporal cortex (Wernicke's area) and frontal cortex (Broca's area). The language areas are bordering one of the major fissures in the brain, the so-called Sylvian fissure. Collectively, this part of the brain is often referred to as perisylvian cortex.] Even today, this model is still influential. For instance, in a recently published study (Moorman et al., 2012) one reads “Broca's area in the frontal lobe and Wernicke's area in the temporal lobe are crucially involved in speech production and perception, respectively.” (p. 12782). Many similar quotations can be found. Despite its impact until this very day, the classical model is wrong (cf. Poeppel et al., 2012). Although Broca's area, Wenicke's area and adjacent cortex are still considered to be key nodes in the language network, the distribution of labor between these regions is different than was claimed in the WLG model. Lesions in Broca's region are since long known to impair not only language production but also language comprehension (Caramazza and Zurif, 1976), whereas lesions in Wernicke's region also affect language production. More recently, neuroimaging studies provided further evidence that the classical view on the role of these regions is no longer tenable. For example, central aspects of language production and comprehension are subserved by shared neural circuitry (Menenti et al., 2011; Segaert et al., 2012). Moreover, the classical model focused on single word processing, whereas a neurobiological account of language processing in its full glory should also take into account what goes on beyond production and comprehension of single words. As a consequence of the mounting evidence against the classical WLG model, in recent years alternative neurobiological models for language have been proposed (e.g., Friederici, 2002; Hagoort, 2005; Hickok and Poeppel, 2007). Here I will focus mainly on the Memory-Unification-Control (MUC) model that I proposed in 2005 (Hagoort, 2005). After describing its three components, I will discuss the evidence that has accumulated in support of the model, and I will suggest extensions of the model on the basis of recent empirical evidence. MEMORY, UNIFICATION, AND CONTROL The MUC model distinguishes three functional components of language processing: Memory, Unification and Control. The Memory component refers to the linguistic knowledge that in the course of language acquisition gets encoded and consolidated in neocortical memory structures. It is the only language-specific component of the model. The knowledge about the building blocks of language (e.g., phonological, morphological, syntactic building blocks) is domain specific and hence coded in a format that is different from, say, color and visual object information. However, language processing is more than memory retrieval and more than the simple concatenation of retrieved lexical items. The expressive power of human language derives from the possibility to combine elements from memory in novel ways. In the model this process of deriving new and complex meaning from the lexical building blocks is referred to as Unification. Unification thus refers to the assembly of pieces stored in memory into larger structures, with contributions from context. Classically, psycholinguistic studies of unification have focused on syntactic analysis. But, crucially, unification operations take place not only at the syntactic processing level, but are a hallmark of language across representational domains (cf. Jackendoff, 2002, 2007). Thus, at the semantic and phonological levels, too, lexical elements are combined and integrated into larger structures. Hence I distinguish between syntactic, semantic and phonological unification (cf. Hagoort, 2005). Finally, the Control component relates language to joint action and social interaction. Executive control is invoked, for instance, when the contextually appropriate target language has to be selected, for handling the joint action aspects of using language in conversational settings, for selecting the appropriate register in different social situations, etcetera. We will later see that languages also have built-in linguistic devices that trigger the attentional control system into operation. In the MUC model, the distribution of labor is as follows (see Figure 2): regions in the temporal cortex (in yellow) and the angular gyrus in parietal cortex subserve the knowledge representations that have been laid down in memory during acquisition. These regions store information including phonological word forms, morphological information, and the syntactic templates associated with noun, verbs, adjectives (for details, see Hagoort, 2003, 2005, 2009a,b). They also include semantic convergence zones, but on the whole conceptual knowledge is quite widely distributed (Binder and Desai, 2011). Dependent on knowledge type, different parts of temporal and parietal cortex are involved. Frontal regions (Broca's area and adjacent cortex; in blue) are crucial for unification operations. These operations generate larger structures from the building blocks that are retrieved from memory. Within left inferior frontal cortex (Unification Space), a spatial activation gradient is observed. The distribution of the activations seems to depend on the type of information that gets unified. Semantic unification recruits BA 47 and BA 45; syntactic unification has its focus in BA 45 and BA 44; phonological processes recruit BA 44 and ventral parts of BA 6 (see Figure 3). In addition, executive control needs to be exerted, such that the correct target language is selected, turn taking in conversation is orchestrated, the correct register is selected, attention is paid to the most relevant information in the input, and so forth. Control regions involve dorsolateral prefrontal cortex (in pink), and midline structure including the anterior cingulate cortex and

Read the complete text