2.2. Planteamientos Teóricos
2.2.6. La televisión como un medio de entretenimiento
Rugg (1990) proposed that because of the sensitivity of the LPC to word frequency, it reflected differences in the 'relative familiarity' of stimuli. Relative familiarity is a concept derived from the dual-process theories of recognition memory (Jacoby, 1983b; Mandler, 1980), and is derived from changes in the efficiency or 'fluency' (Jacoby, 1983b) with which words are perceived after repetition. Thus this later component was held to reflect the operation of processes which allowed an event to be judged as old in the absence of episodic information concerning its specific occurrence.
In contrast to the 'familiarity' account of the LPC proposed by Rugg (1990), Van Petten et al (1991) suggested that the LPC reflected episodic retrieval processes. They argued that the differences between the negative-going LPC repetition effect which they observed and the positive-going LPC repetition effect observed by Rugg (1990) reflected the varying demands on episodic memory imposed by the differing experimental tasks. Van Petten et al (1991) argued that in a task where only words repeat, the LDT reduces, at least in part, to one of episodic memory for repeating stimuli. Under this task description, the demands on episodic memory are primarily made when a word is repeated. Thus if the LPC reflects the operation of episodic memory then second presentations will be more positive going than first presentations - the findings reported by Rugg (1990).
In contrast, Van Petten et al (1991) argue that in their text reading experiment, the primary demands on episodic memory are made when a word is encountered for the first time since it is then that the meaning of the word will be retrieved. When a word is repeated in a passage of text, the meaning is more likely to still be in working memory, and therefore the likelihood of retrieval from episodic memory is less. Thus episodic retrieval processes will be greater on the first presentation and repetition will result in a negative-going shift in the LPC - the results observed by Van Petten et al (1991).
Besson and Kutas (1993) also advocated an episodic account of the LPC in which this
component reflected the degree to which a sentence terminal word had been 'remembered', and more recently this account has gained support from the results of investigations of the
demonstrated that there was a large positive-going shift in the ERP in the region of the LPC only when subjects could correctly categorise a word as being new or repeated and, for repeated words, could also identify whether they had been seen or heard on their first presentation. That modulations of the LPC were oberved only when such contextual
information was available suggests this component reflects recollection of specific episodes because only recollection is thought to supply information about the context in which an item was encountered (see also Palier, Kutas, & Mclsaac, 1995; Smith, 1993).
2,5 Modulations of the N400 and their interpretation
There are a number of reviews which have been mainly concerned with the functional
significance of the N400 (Fischler & Raney, 1991; Holcomb & Osterhout, 1995; Kutas & Van Petten, 1988), but I will briefly describe the initial demonstrations of modulations of the N400 component and two opposing accounts of its functional role.
The modulation of the N400 component was first demonstrated by Kutas and Hillyard (1980; 1984). They presented subjects with sentence frames such as:
'He spread the warm bread with ...'
These sentence frames were completed with words which were either semantically congruous (e.g. 'butter') or semantically incongruous (e.g. 'socks') with the frame, or with words which were physically incongruous (presented in a different sized font). Kutas and Hillyard had predicted that in the conditions in which the sentence frame was completed by an unexpected stimulus they would see a modulation of the P3 component.
In the case of the semantically incongruent sentence ending only, rather than a modulation of the P3, Kutas and Hillyard (1980) observed a negative-going deflection of the waveform which peaked at around 400 ms. In the light of the apparent specificity of this N400 component to semantic manipulations, Kutas and Hillyard (1980) proposed that it reflected an interruption to the normal processing of meaning which occurs during the reading of a sentence. This
'semantic hypothesis' received some support with a further demonstration of the sensitivity of the region of the waveform around 400 ms to semantic, but not grammatical, incongruency (Kutas & Hillyard, 1983, but see; Muente, Heinze, & Mangun, 1993, and Rosier, Putz, Friederici, & Hahne, 1993).
Kutas and Hillyard (1984) presented subjects with sentence frames which were completed by words which varied in 'cloze probability' (Taylor, 1953). Cloze probability describes the likelihood that, given a particular sentence frame, a subject would complete the sentence with a particular word. The more likely a particular word would be used to complete a sentence the higher its cloze probability. Kutas and Hillyard (1984) demonstrated that the amplitude of the
N400 varies inversely with cloze probability and argued that the amplitude of the N400 was determined by semantic 'expectancy' rather than incongruity
per se.
Critically however, Kutas and Hillyard (1984) also demonstrated that the amplitude of the N400 was attenuated not only by words which were of high cloze probability, but also by words which, although themselves of low cloze probability, were semantically related to high cloze probability words. Thus the sentence frame
He liked lemon and sugar in his ...'
could be completed not only by its best completion (i.e. 'tea') but also by a semantic associate of the best completiton (e.g. 'coffee') and an attenuation of the amplitude of the N400 would result. Kutas and Hillyard (1984) suggested that the N400 reflected a spread of activation between semantically related words, and that the N400 was an electrophysiological indicator of automatic lexical and/or semantic processing and was specific to linguistic or potentially linguistic materials (see also Kutas, Lindamood, & Hillyard, 1984). Consistent with this proposal, Besson and Macar (1987) showed that an N400 was not generated when any
expectation of a forthcoming experimental item was not realized. They presented subjects with simple musical scales or simple tunes, and recorded ERPs to the final note which either did or did not complete the scale or tune appropriately. In these circumstances neither an N400 nor a 'congruity' effect was observed.
Van Petten and Kutas (1990) reported that the effects of semantic congruity and frequency had interactive effects on the amplitude of N400, whilst Van Petten et al (1991) argued that frequency and repetition have similar effects on the amplitude of the N400 as would be expected if they exerted their effects through a common mechanism. Besson et al (1992) suggest that the interaction between congruity and repetition 'can be interpreted in like manner' (p. 144) to the interaction between repetition and frequency; that is they reflect a process of lexical identification. Van Petten et al (1991) argued that 'Our knowledge of the N400 to date characterises it as a rather ubiquitous marker of lexical processing' (p. 140). Besson and Kutas (1993) also argued that the modulations of the N400 reflect 'language-specific knowledge', whilst Holcomb and Neville (1990) have argued that the amplitude of the N400 elicited by a word reflects the degree to which a lexical representation of that word is activated prior to its presentation.
This ‘lexical’ hypothesis can be contrasted with a 'contextual integration' hypothsis first proposed by Halgren and Smith (1987) and developed by Rugg (1990, see also Rugg & Doyle,
1994), Under this hypothesis modulations of the N400 reflect the ease or extent to which the representation of an item can be integrated into a representation of the context in which it occured. N400 amplitude is modulated by any variable that affects this process of integration. Repetition effects arise not because there is a change in the representation of a particular item
but because the contextual integration of an item is easier, or needs to be done to a lesser extent, as a result of its prior presentation.
The relative merits of these two different accounts of N400 effects will form the backdrop to the discussion of studies which have investigated the way in which N400 and repetition effects can be modulated in response to a variety of experimental variables. I will begin by describing some of the studies which have investigated the sensitivity of the N400 component to semantic priming manipulations.