N400 (neuroscience)

The N400 is a component of time-locked

The N400 was first discovered by Marta Kutas and Steven Hillyard in 1980.^[4] They conducted the first experiment looking at the response to unexpected words in read sentences, expecting to elicit a P300 component. The P300 had previously been shown to be elicited by unexpected stimuli. Kutas and Hillyard therefore used sentences with anomalous endings (i.e.I take coffee with cream and dog), expecting to see a P300 to the unexpected sentence-final words. However, instead of eliciting a large positivity, these anomalous endings elicited a large negativity, relative to the sentences with expected endings (i.e. He returned the book to the library) In the same paper, they confirmed that the negativity was not just caused by any unexpected event at the end of a sentence, since a semantically expected but physically unexpected word (i.e. She put on her high-heeled SHOES) elicited a P300 instead of negativity in the N400 window. This finding showed that the N400 is related to semantic processing, and is not just a response to unexpected words.

Component characteristics

The N400 is characterized by a distinct pattern of electrical activity that can be observed at the scalp. As its name indicates, this waveform peaks around 400 ms post-stimulus onset, with negativity that can be observed in the time window ranging from 250-500 ms. This latency (delay between stimulus and response) is very stable across tasks—little else besides

An example of an experimental task used to study the N400 is a priming paradigm. Subjects are shown a list of words in which a prime word is either associatively related to a target word (e.g. bee and honey), semantically related (e.g. sugar and honey) or a direct repetition (e.g. honey and honey). The N400 amplitude seen to the target word (honey) will be reduced upon repetition due to semantic priming.^[1] The amount of reduction in amplitude can be used to measure the degree of relatedness between the words.

Another widely used experimental task used to study the N400 is sentence

As previously mentioned, the N400 response is seen to all meaningful, or potentially meaningful, stimuli. As such, a wide range of paradigms can be used to study it. Experiments involving the presentation of spoken words,^[5] acronyms,^[6] pictures embedded at the end of sentences,^[7] music,^[8] words related to current context or orientation^[9] and videos of real-word events,^[10] have all been used to study the N400, just to name a few.

Functional sensitivity

Extensive research has been carried out to better understand what kinds of experimental manipulations do and do not affect the N400. General findings are discussed below.

Factors that affect N400 amplitude

The frequency of a word's usage is known to affect the amplitude of the N400. With all else being constant, highly frequent words elicit reduced N400s relative to infrequent words.^[11] As previously mentioned, N400 amplitude is also reduced by repetition, such that a word's second presentation exhibits a more positive response when repeated in context.^[12] These findings suggest that when a word is highly frequent or has recently appeared in context, it eases the semantic processing thought to be indexed by the N400, reducing its amplitude.

N400 amplitude is also sensitive to a word's orthographic neighborhood size, or how many other words differ from it by only one letter (e.g. boot and boat). Words with large neighborhoods (that have many other physically similar items) elicit larger N400 amplitudes than do words with small neighborhoods.^[13] This finding also holds true for pseudowords, or pronounceable letter strings that are not real words (e.g. flom), which are not themselves meaningful but look like words. This has been taken as evidence that the N400 reflects general activation in the comprehension network, such that an item that looks like many words (regardless of whether it itself is a word) partially activates the representations of similar-looking words, producing a more negative N400.

The N400 is sensitive to priming: in other words, its amplitude is reduced when a target word is preceded by a word that is semantically, morphologically, or orthographically related to it.^[1]

In a sentence context, an important determinant of N400 amplitude elicited by a word is its cloze probability. Cloze probability is defined as the probability of the target word completing that particular sentence frame. Kutas and Hillyard (1984) found that a word's N400 amplitude has a nearly inverse linear relationship with its cloze probability.^[14] That is, as a word becomes less expected in context, its N400 amplitude is increased relative to more expected words. Words that are incongruent with a context (and thus have a cloze probability of 0) elicit large N400 amplitudes as well (although the amplitude of the N400 for incongruent words is also modulated by the cloze probability of the congruent word that would have been expected in its place^[15] Relatedly, the N400 amplitude elicited by open-class words (i.e. nouns, verbs, adjectives, and adverbs) is reduced for words appearing later in a sentence compared to earlier words.^[11] Taken together, these findings suggest that when the prior context builds up meaning, it makes the processing of upcoming words that fit with that context easier, reducing the N400 amplitude they elicit.

Factors that do not affect N400 amplitude

While the N400 is larger to unexpected items at the end of a sentence, its amplitude is generally unaffected by

Additionally, grammatical violations do not elicit a large N400 response. Rather, these types of violations show a large positivity from about 500-1000 ms after stimulus onset, known as the P600.^[2]

Factors that affect N400 latency

A striking feature of the N400 is the general invariance of its peak latency. Although many different experimental manipulations affect the amplitude of the N400, few factors (aging and disease states and language proficiency being rare examples) alter the time it takes for the N400 component to reach a peak amplitude.^[19]

Sources

Although localization of the neural generators of an ERP signal is difficult due to the spreading of current from the source to the sensors, multiple techniques can be used to provide converging evidence about possible neural sources.^[20] Using methods such as recordings directly off the surface of the brain or from electrodes implanted in the brain, evidence from brain damaged patients, and magnetoencephalographic (MEG) recordings (which measure magnetic activity at the scalp associated with the electrical signal measured by ERPs), the left temporal lobe has been highlighted as an important source for the N400, with additional contributions from the right temporal lobe.^[21] More generally, however, activity in a wide network of brain areas is elicited in the N400 time window, suggesting a highly distributed neural source.^[2]

Theories

There is still much debate as to exactly what kind of neural and comprehension processes the N400 indexes. Some researchers believe that the underlying processes reflected in the N400 occur after a stimulus has been recognized. For example, Brown and Hagoort (1993) believe that the N400 occurs late in the processing stream, and reflects the integration of a word's meaning into the preceding context (see Kutas & Federmeier, 2011,

More recent accounts posit that the N400 represents a broader range of processes indexing access to semantic memory. According to this account, it represents the binding of information obtained from stimulus input with representations from short- and long-term memory (such as recent context, and accessing a word's meaning in long term memory) that work together to create meaning from the information available in the current context (Federmeier & Laszlo, 2009; see Kutas & Federmeier, 2011^[2]).

Another account is that the N400 reflects prediction error or surprisal. Word-based surprisal was a strong predictor of N400 amplitude in an ERP corpus.^[23] In addition, connectionist models make use of prediction error for learning and linguistic adaptation, and these models can explain several N400/P600 results in terms of prediction error propagation for learning.^[24]

It may also be that the N400 reflects a combination of these or other factors. Nieuwland et al. (2019) argue that the N400 is actually made up of two sub-components, with predictability affecting the early part of the N400 (200-500 ms after stimulus onset) and plausibility affecting it later (350-650 ms after stimulus onset).^[25] This suggests that the N400 reflects both access to semantic memory (which is sensitive to prediction), and semantic integration (sensitive to plausibility).

As research in the field of electrophysiology continues to progress, these theories will likely be refined to include a complete account of just what the N400 represents.

See also

References