Refine
Year of publication
Document Type
- Article (11)
- Postprint (5)
- Doctoral Thesis (1)
Language
- English (17) (remove)
Keywords
- N400 (17) (remove)
Institute
Prediction is often regarded as a central and domain-general aspect of cognition. This proposal extends to language, where predictive processing might enable the comprehension of rapidly unfolding input by anticipating upcoming words or their semantic features. To make these predictions, the brain needs to form a representation of the predictive patterns in the environment. Predictive processing theories suggest a continuous learning process that is driven by prediction errors, but much is still to be learned about this mechanism in language comprehension. This thesis therefore combined three electroencephalography (EEG) experiments to explore the relationship between prediction and implicit learning at the level of meaning.
Results from Study 1 support the assumption that the brain constantly infers und updates probabilistic representations of the semantic context, potentially across multiple levels of complexity. N400 and P600 brain potentials could be predicted by semantic surprise based on a probabilistic estimate of previous exposure and a more complex probability representation, respectively.
Subsequent work investigated the influence of prediction errors on the update of semantic predictions during sentence comprehension. In line with error-based learning, unexpected sentence continuations in Study 2 ¬– characterized by large N400 amplitudes ¬– were associated with increased implicit memory compared to expected continuations. Further, Study 3 indicates that prediction errors not only strengthen the representation of the unexpected word, but also update specific predictions made from the respective sentence context. The study additionally provides initial evidence that the amount of unpredicted information as reflected in N400 amplitudes drives this update of predictions, irrespective of the strength of the original incorrect prediction.
Together, these results support a central assumption of predictive processing theories: A probabilistic predictive representation at the level of meaning that is updated by prediction errors. They further propose the N400 ERP component as a possible learning signal. The results also emphasize the need for further research regarding the role of the late positive ERP components in error-based learning. The continuous error-based adaptation described in this thesis allows the brain to improve its predictive representation with the aim to make better predictions in the future.
Classical linguistic theory assumes that formal aspects, like sound, are not internally related to the meaning of words. However, recent research suggests language might code affective meaning such as threat and alert sublexically. Positing affective phonological iconicity as a systematic organization principle of the German lexicon, we calculated sublexical affective values for sub-syllabic phonological word segments from a large-scale affective lexical German database by averaging valence and arousal ratings of all words any phonological segment appears in. We tested word stimuli with either consistent or inconsistent mappings between lexical affective meaning and sublexical affective values (negative-valence/high-arousal vs. neutral-valence/lowarousal) in an EEG visual-lexical-decision task. A mismatch between sublexical and lexical affective values elicited an increased N400 response. These results reveal that systematic affective phonological iconicity - extracted from the lexicon - impacts the extraction of lexical word meaning during reading.
Intuitively, strongly constraining contexts should lead to stronger probabilistic representations of sentences in memory. Encountering unexpected words could therefore be expected to trigger costlier shifts in these representations than expected words. However, psycholinguistic measures commonly used to study probabilistic processing, such as the N400 event-related potential (ERP) component, are sensitive to word predictability but not to contextual constraint. Some research suggests that constraint-related processing cost may be measurable via an ERP positivity following the N400, known as the anterior post-N400 positivity (PNP). The PNP is argued to reflect update of a sentence representation and to be distinct from the posterior P600, which reflects conflict detection and reanalysis. However, constraint-related PNP findings are inconsistent. We sought to conceptually replicate Federmeier et al. (2007) and Kuperberg et al. (2020), who observed that the PNP, but not the N400 or the P600, was affected by constraint at unexpected but plausible words. Using a pre-registered design and statistical approach maximising power, we demonstrated a dissociated effect of predictability and constraint: strong evidence for predictability but not constraint in the N400 window, and strong evidence for constraint but not predictability in the later window. However, the constraint effect was consistent with a P600 and not a PNP, suggesting increased conflict between a strong representation and unexpected input rather than greater update of the representation. We conclude that either a simple strong/weak constraint design is not always sufficient to elicit the PNP, or that previous PNP constraint findings could be an artifact of smaller sample size.
Intuitively, strongly constraining contexts should lead to stronger probabilistic representations of sentences in memory. Encountering unexpected words could therefore be expected to trigger costlier shifts in these representations than expected words. However, psycholinguistic measures commonly used to study probabilistic processing, such as the N400 event-related potential (ERP) component, are sensitive to word predictability but not to contextual constraint. Some research suggests that constraint-related processing cost may be measurable via an ERP positivity following the N400, known as the anterior post-N400 positivity (PNP). The PNP is argued to reflect update of a sentence representation and to be distinct from the posterior P600, which reflects conflict detection and reanalysis. However, constraint-related PNP findings are inconsistent. We sought to conceptually replicate Federmeier et al. (2007) and Kuperberg et al. (2020), who observed that the PNP, but not the N400 or the P600, was affected by constraint at unexpected but plausible words. Using a pre-registered design and statistical approach maximising power, we demonstrated a dissociated effect of predictability and constraint: strong evidence for predictability but not constraint in the N400 window, and strong evidence for constraint but not predictability in the later window. However, the constraint effect was consistent with a P600 and not a PNP, suggesting increased conflict between a strong representation and unexpected input rather than greater update of the representation. We conclude that either a simple strong/weak constraint design is not always sufficient to elicit the PNP, or that previous PNP constraint findings could be an artifact of smaller sample size.
We argue that natural language can be usefully described as quasi-compositional and we suggest that deep learning-based neural language models bear long-term promise to capture how language conveys meaning. We also note that a successful account of human language processing should explain both the outcome of the comprehension process and the continuous internal processes underlying this performance. These points motivate our discussion of a neural network model of sentence comprehension, the Sentence Gestalt model, which we have used to account for the N400 component of the event-related brain potential (ERP), which tracks meaning processing as it happens in real time. The model, which shares features with recent deep learning-based language models, simulates N400 amplitude as the automatic update of a probabilistic representation of the situation or event described by the sentence, corresponding to a temporal difference learning signal at the level of meaning. We suggest that this process happens relatively automatically, and that sometimes a more-controlled attention-dependent process is necessary for successful comprehension, which may be reflected in the subsequent P600 ERP component. We relate this account to current deep learning models as well as classic linguistic theory, and use it to illustrate a domain general perspective on some specific linguistic operations postulated based on compositional analyses of natural language. This article is part of the theme issue 'Towards mechanistic models of meaning composition'.
Increased N400 amplitudes on indefinite articles (a/an) incompatible with expected nouns have been initially taken as strong evidence for probabilistic pre-activation of phonological word forms, and recently been intensely debated because they have been difficult to replicate. Here, these effects are simulated using a neural network model of sentence comprehension that we previously used to simulate a broad range of empirical N400 effects. The model produces the effects when the cue validity of the articles concerning upcoming noun meaning in the learning environment is high, but fails to produce the effects when the cue validity of the articles is low due to adjectives presented between articles and nouns during training. These simulations provide insight into one of the factors potentially contributing to the small size of the effects in empirical studies and generate predictions for cross-linguistic differences in article induced N400 effects based on articles’ cue validity. The model accounts for article induced N400 effects without assuming pre-activation of word forms, and instead simulates these effects as the stimulus-induced change in a probabilistic representation of meaning corresponding to an implicit semantic prediction error.
Increased N400 amplitudes on indefinite articles (a/an) incompatible with expected nouns have been initially taken as strong evidence for probabilistic pre-activation of phonological word forms, and recently been intensely debated because they have been difficult to replicate. Here, these effects are simulated using a neural network model of sentence comprehension that we previously used to simulate a broad range of empirical N400 effects. The model produces the effects when the cue validity of the articles concerning upcoming noun meaning in the learning environment is high, but fails to produce the effects when the cue validity of the articles is low due to adjectives presented between articles and nouns during training. These simulations provide insight into one of the factors potentially contributing to the small size of the effects in empirical studies and generate predictions for cross-linguistic differences in article induced N400 effects based on articles’ cue validity. The model accounts for article induced N400 effects without assuming pre-activation of word forms, and instead simulates these effects as the stimulus-induced change in a probabilistic representation of meaning corresponding to an implicit semantic prediction error.
Increased N400 amplitudes on indefinite articles (a/an) incompatible with expected nouns have been initially taken as strong evidence for probabilistic pre-activation of phonological word forms, and recently been intensely debated because they have been difficult to replicate. Here, these effects are simulated using a neural network model of sentence comprehension that we previously used to simulate a broad range of empirical N400 effects. The model produces the effects when the cue validity of the articles concerning upcoming noun meaning in the learning environment is high, but fails to produce the effects when the cue validity of the articles is low due to adjectives presented between articles and nouns during training. These simulations provide insight into one of the factors potentially contributing to the small size of the effects in empirical studies and generate predictions for cross-linguistic differences in article induced N400 effects based on articles’ cue validity. The model accounts for article induced N400 effects without assuming pre-activation of word forms, and instead simulates these effects as the stimulus-induced change in a probabilistic representation of meaning corresponding to an implicit semantic prediction error.
Previous cross-modal priming studies showed that lexical decisions to words after a pronoun were facilitated when these words were semantically related to the pronoun's antecedent. These studies suggested that semantic priming effectively measured antecedent retrieval during coreference. We examined whether these effects extended to implicit reading comprehension using the N400 response. The results of three experiments did not yield strong evidence of semantic facilitation due to coreference. Further, the comparison with two additional experiments showed that N400 facilitation effects were reduced in sentences (vs. word pair paradigms) and were modulated by the case morphology of the prime word. We propose that priming effects in cross-modal experiments may have resulted from task-related strategies. More generally, the impact of sentence context and morphological information on priming effects suggests that they may depend on the extent to which the upcoming input is predicted, rather than automatic spreading activation between semantically related words.
Scripts that store knowledge of everyday events are fundamentally important for managing daily routines. Content event knowledge (i.e., knowledge about which events belong to a script) and temporal event knowledge (i.e., knowledge about the chronological order of events in a script) constitute qualitatively different forms of knowledge. However, there is limited information about each distinct process and the time course involved in accessing content and temporal event knowledge. Therefore, we analyzed event-related potentials (ERPs) in response to either correctly presented event sequences or event sequences that contained a content or temporal error. We found an N400, which was followed by a posteriorly distributed P600 in response to content errors in event sequences. By contrast, we did not find an N400 but an anteriorly distributed P600 in response to temporal errors in event sequences. Thus, the N400 seems to be elicited as a response to a general mismatch between an event and the established event model. We assume that the expectancy violation of content event knowledge, as indicated by the N400, induces the collapse of the established event model, a process indicated by the posterior P600. The expectancy violation of temporal event knowledge is assumed to induce an attempt to reorganize the event model in working memory, a process indicated by the frontal P600. (C) 2015 Elsevier Ltd. All rights reserved.