The neurocognitive processing of plausibility and real-world knowledge:A cross-linguistic investigation
Our knowledge about concepts and meanings is at the very heart of human cognition. In everyday life, we have to interact with our environment in a variety of different ways. Our actions are guided by what we know and believe about the world and this knowledge derives primarily from previous senso...
|PDF Full Text
No Tags, Be the first to tag this record!
|Our knowledge about concepts and meanings is at the very heart of human cognition.
In everyday life, we have to interact with our environment in a variety of different
ways. Our actions are guided by what we know and believe about the world and this
knowledge derives primarily from previous sensory and perceptual experiences. The
fact that we are capable of engaging with our environment in an appropriate and efficient
way means that we have learnt (how) to make sense of the events and entities we
are faced with in day-to-day life. We are thus able to recognise and name both physical
objects and abstract concepts, to categorise and associate them based on their specific
properties, to interpret other people’s intentions, and to judge cause and effect of
their actions as well as our own. Moreover, the ability to represent this wealth of
knowledge about the real world in the conceptualised and symbolic form of language
is believed to be exclusive to humans. Our language capacity allows us to communicate
with others about past and future events or to describe fictitious scenarios by
combining previously acquired concepts in a novel way without the need for external
stimulation. Thus language forms a primary means of interacting with those around us
by allowing us to express our own thoughts and comprehend those of others. As long
as language processing proceeds in an undisturbed manner, we are largely unaware of
the underlying mechanisms that support the seemingly effortless interpretation of linguistic
input. The importance of these processes for successful communication, however,
becomes all the more apparent when language processing is disrupted, for example,
by brain lesions that render semantic analysis difficult or impossible.
Scientific research that aims to uncover and define cognitive or neural mechanisms
underlying semantic processing is inevitably faced with the complexity and
wealth of semantic relationships that need to be taken into account. In absence of noninvasive
neurocognitive methods and insights gleaned from modern neurobiology,
early research had a limited impact on our understanding of how semantic processing
is implemented in the human brain. Traditional neurological models of language have
been based primarily on lesion-deficit data, and thus supported the view that certain
areas of the brain were exclusively dedicated to the processing of language-specific
functions (Geschwind, 1970; Lichtheim, 1885; Wernicke, 1874). Furthermore, classical
theories of sensory processing viewed the brain as a purely stimulus-driven system that retrieves and combines individual low-level aspects or features in an automated,
passive and context-independent manner (Biederman, 1987; Burton & Sinclair, 1996;
Hubel & Wiesel, 1965; Massaro, 1998).
After a recent paradigm shift in the cognitive neurosciences, current theories
of sensory processing are now based on the concept of the brain as a highly active,
adaptive and dynamic device. In this sense, language comprehension, like many other
higher-cognitive functions, is shaped by a flexible interaction of a number of different
processes and information sources that include so-called bottom-up signals, i.e., the
actual sensory input and processes related to their forward propagation, and top-down
processes that generate predictions and expectations based on prior experience and
perceived probabilities. Therefore, accounts that view semantic processing as a dynamic
and active construction of meaning that is highly sensitive to contextual influences
seem most probable from a neurobiological perspective. Results from electrophysiological
and neuroimaging research on semantic analysis in sentence and discourse
context have provided evidence for top-down influences from the very beginning.
In addition, recent ERP results have suggested that the interaction between topdown
and bottom-up information is more flexible and dynamic than previously assumed.
Yet, the importance of predictions and expectations has long been neglected in
models of semantic processing and language comprehension in general.
Neuroimaging data have provided us with a long list of brain regions that have
been implicated in different aspects of semantic analysis. We are only beginning to
understand the role(s) that these regions play and how they interact to support the
flexible and efficient construction of meaning.
The aim of the present thesis is to gain a more comprehensive view on the
computational mechanisms underlying language processing by investigating how bottom-
up and top-down information and processes interactively contribute to the semantic
analysis in sentences and discourse. To this end, we conducted a total of five studies
that used either event-related potentials or functional neuroimaging to shed light
on this matter from different perspectives.
The thesis is divided into two main parts: Part I (chapters 1-5) provides an
overview on previous results from electrophysiology and neuroimaging on semantic
processing as well as a description and discussion of the studies conducted in the present thesis. Part II (chapters 6-9) consists of three research articles that describe
and discuss the results of five experimental studies.
In Part I, Chapter 2 gives a brief introduction to the event-related potential and
functional neuroimaging techniques and reviews the most relevant results and theories
that have emerged from studies on sentence and discourse processing. Chapter 3 highlights
the research questions targeted in each of the experimental studies and describes
and discusses the most relevant findings against the background established by Chapter
2. Chapters 4 and 5 conclude Part I by placing the presented results in a broader
context and by briefly outlining future directions.
Part II begins with a survey of the three studies reported in the subsequent
chapters. Chapter 7 highlights the results of the first study, a German ERP experiment
that investigated the impact of capitalisation, i.e., a purely form-based and contextually
independent bottom-up manipulation, on the processing of semantic anomalies
in single sentences. Chapter 8 comprises three ERP experiments that used both easy
and hard to detect semantic anomalies in German and English to corroborate the assumption
that the weighting of top-down and bottom-up information cues might be
determined in a language-specific way. Chapter 9, the final chapter of the thesis, describes
and discusses the results of the third study, in which the impact of embedding
context on the required depth of semantic processing was examined using functional