Modulation of world knowledge comprehension by discourse

Modulation of world knowledge comprehension by discourse
an fMRI study
Laura Menenti1, Karl Magnus Petersson1,2, René Scheeringa1 & Peter Hagoort1,2
1F.C.Donders
Centre for Cognitive Neuroimaging; 2Max Planck Institute for Psycholinguistics
Nijmegen, the Netherlands
Introduction
Results
In the Memory, Unification and Control model (Hagoort, 2005) the Left
Inferior Frontal Gyrus (LIFG) is hypothesized to unify information from
different sources into a coherent representation of a multi-word utterance.
Hagoort et al. (2004) showed that integration of both meaning and world
knowledge takes place in the LIFG, in BA 47 and 45.
In an ERP study Hald et al. (in press) found a modulation of the N400
effect of world knowledge violations using local contexts (which made
world knowledge violations more acceptable), suggesting facilitation of
unification.
We used a similar paradigm in fMRI, to study whether a facilitation of
unification also occurs in the LIFG. We hypothesized that the effect of
world knowledge violations in left BA 45 and 47 would decrease when
preceded by a local context. This would be consistent with the view that
the LIFG integrates information from different sources.
Since discourse processing tends to engage the right hemisphere we also
tested for the same effects in right BA 45 and 47.
*
*
*
*
Figure 3: Results of ROI analyses for bilateral BA 45 and 47. An interaction
between context and WK was found in the RH, but not the LH.
Planned comparisons in the four ROIs revealed that in regions the effect of
world knowledge violations was only found to be significant for the neutral
context. However, the interaction was only significant in the RH, where a
cross-over pattern was found.
Whole-brain analyses revealed that the Left Angular gyrus also showed a
significant cross-over interaction of context by world knowledge.
ND-WKV
LD-WKV
LD-COR
ND-COR
Figure 1: WK-effect in
Hagoort et al. (2004)
Figure 2: Modulation of N400-effect of WK-violations
by context in Hald et al. (in press)
Method
Materials
132 stimulussets in Dutch, consisting of two tightly matched contexts and
two critical sentences, differing in one word.
Neutral context: Long-distance transport is growing customary. Most
passengers are transported by plane. Planes have grown bigger and
faster and the journey keeps getting shorter.
Local context: Long-distance transport is growing customary. Most cargo
is transported by ship. Ships have grown bigger and faster and the
journey keeps getting shorter.
Critical sentence: The journey from Amsterdam to Washington takes
about six hours/weeks and can be expensive.
Experiment
32 native Dutch speakers participated in the experiment. They read the
materials silently and responded to comprehension questions on 10% of
the trials. 28 slices with a voxel size of 3.5mm3 were acquired with a TR of
1.86s on a Siemens 3T MR-scanner.
Figure 4: Cross-over interaction of context x
world knowledge in LAG
Conclusion
As predicted, world knowledge violations elicited a stronger response than
correct sentences in L BA45 and BA47, replicating Hagoort et al.
(2004).The same effects were also found in the right hemisphere
homotopic regions. However, when presented after the local context the
effect of world knowledge was no longer found in any of the ROIs.
The cross-over interaction in R BA45 and BA47 suggests that these areas
are relatively more sensitive to the fit between the critical sentence and its
preceding context. This is consistent with the idea that discourse
processing tends to engage the right- relatively more than the left
hemisphere. It therefore appears that not only the LIFG but also the RIFG
is involved in unifying information from different sources (in this case,
information in long-term memory and ongoing discourse) into a coherent
representation (cf., Forkstam et al., 2006).
A complete cross-over interaction was found in the left angular gyrus,
indicating that this area is only sensitive to the fit between the context and
the critical sentence. This may be indicative of top-down influences on
word recognition.
Contact: [email protected]
References
Forkstam, C., Hagoort, P., Fernandez, G., Ingvar, M., Petersson, K. M., 2006. Neural correlates of artificial
syntactic structure classification. NeuroImage 32: 956-967.
Hagoort, P. (2005) On Broca, brain, and binding: a new framework. Trends in Cognitive Sciences, 9, 416-423.
Hagoort, P., Hald, L., Bastiaansen, M., & Petersson, K.M. (2004) Integration of word meaning and world
knowledge in language comprehension. Science, 304, 438-441.
Hald, L., Steenbeek-Planting, E, & Hagoort, P. (in press). The interaction of discourse context and world
knowledge in online sentence comprehension. Evidence from the N400. Brain Research, special issue on
Discourse Comprehension.