Research reportEvent related brain potentials and illusory memories: the effects of differential encoding
Introduction
The act of remembering is the outcome of multiple, fundamentally reconstructive, component processes (for an overview see [51], [61]). These processes include inferences drawn on the basis of feelings of familiarity elicited by a stimulus, as well as the active recollection of a memory trace [30], [66]. Although the majority of memory studies have focused on whether or not studied items are accurately recalled or recognized, more recently the investigation of memory illusions, so-called false memories, has received increased attention. In the typical laboratory study of false memory, participants learn lists of associate words of a non presented word, the so-called LURE word. The critical finding, replicated many times [33], [45], [50], is that in a subsequent recall or recognition test, participants falsely recall or recognize the LURE words at a much higher rate than words unrelated to the study lists1.
One explanation of false recognition holds that it is due to a feeling of familiarity and is not due to the active recollection of a memory trace. According to this explanation, LURE words feel familiar and are judged old because they are broadly consistent with the conceptual features that were studied; they largely match the overall themes of words encountered in the study phase [61], [60]2. Support for this view comes from studies showing that more sensory and distinctive details (item specific memory traces) are retrieved for true than for false recognition ( [31], [34], [40], cf. [60]).
An alternative explanation suggests that the false memory phenomena is based on both inferences drawn on the basis of feelings of familiarity and the active recollection of a memory trace. According to this model, the non studied LURE words are activated, and hence memory traces formed, during study of the associated words via spreading activation through the mental lexicon [5], [69]. For example, studying words like butter or sandwich could lead to the activation of the word bread. In the test phase, participants may correctly recognize butter as a studied word but may falsely recognize a LURE word like bread because it was also activated during study. Consequently, in this model, false recognition results from feelings of familiarity that arise due to conceptual similarities between OLD and LURE words and from prior activation in the study phase with a failure to attribute that activation to its correct source [15], [25], [23]. Support for this view is provided by Roediger and McDermott [52]. They required participants to indicate whether an old response was based on consciously recollected aspects of prior experience, i.e. a memory trace (‘Remember’ response), or merely on the belief that a test word had occurred in study without any recollection of the specific study episode, i.e. familiarity (‘Know’ response). Importantly, ‘Remember’ response rates following true and false recognition were equal, indicating that participants used similar information for true and false recognition (for similar results see Ref. [45]).
Event-related potentials (ERPs) can provide an additional source of information about whether the cognitive processes involved in true and false recognition are the same or different. This is because the timing and scalp topography of particular ERP-components allows inferences about the timing and nature of cognitive processes underlying recognition memory judgments to be made [10], [26], [55]. In the case of studies of false recognition the reasoning is that if the same cognitive processes underlie true and false recognition, then the ERP patterns elicited should be the same. Indeed, Johnson et al. [24] and Düzel et al. [12] found equivalent ERPs for true and false recognition when a random word order test presentation was used, suggesting that true and false recognition engage the same neural and cognitive processes. Assuming that both familiarity and active recollection processes underlie true recognition in their experiments, then these results are consistent with the view that both processes also underlie false recognition.
However, both of these ERP studies used word lists from the false memory paradigm introduced by Deese [9]. In this paradigm, the LURE words are theme words (e.g. sweet) that are more highly associated with the studied words (e.g. candy, sour) than the studied words are to each other (for a critical discussion of the Deese paradigm see Refs. [39], [53]). Consequently, a difficulty with this paradigm is that it may enhance activation of LURE words via associative mechanisms in the encoding phase, leading to equivalent activation for LURE and studied words. This, in turn, could result in equivalent ERP waveforms for true and false recognition. Given this possibility, the goal of the present study was to determine whether OLD and LURE words elicited equivalent ERPs when the LURE words were equivalently related to a studied theme as were the OLD words. As the ERP data were analysed within a theoretical framework based on ERP studies of true recognition a brief description of the evidence for this framework follows.
True recognition elicits more positive ERP waveforms than correctly rejected NEW words in explicit old/new recognition tests (for reviews see Refs. [27], [54]). These ERP old/new effects have a broad temporal and spatial distribution and can be decomposed into at least three spatio-temporally specific effects [13], [36] that are associated with distinct cognitive processes underlying true recognition. These effects are an early fronto-medial old/new ERP effect, a somewhat later parietal old/new ERP effect and a late right frontal old/new ERP effect.
The fronto-medial focused positivity starts around 300 ms and lasts approximately 200 ms. This early old/new effect is assumed to arise from the attenuation of a frontally focused N400-like component that occurs because access to conceptual and perceptual information related to the test word is facilitated [6], [7], [28], [35], [37], [46] and results in a feeling of familiarity [36]. That the effect is not driven by active recollection of item specific information is supported by its insensitivity to depth of processing manipulations [57].
The second positive deflection, maximal at parietal locations, starts around 400 ms and lasts for several hundred milliseconds. This parietal old/new ERP-effect, which is usually left lateralised or bilateral shows larger amplitudes for deeply than for shallowly encoded items [43], [68]. Consequently, a relation to consciously controlled recollection of item specific information from the study phase is assumed ([12], [42], [65], [71] and see Refs. [27], [54] for reviews).
Third, a late right frontal old/new effect, which onsets around 800 ms, but is sustained longer in time than the ERP-effects described above, has also been reported [2], [37], [71]. At present there is no consensus on its precise functional significance [13], [36]. Although a relation to recognition related processes is assumed, sometimes the effect is present for NEW words [49], suggesting that the late right frontal positivity also reflects general task related processes [11].
In the present study, if false recognition is based on both familiarity and active recollection, then the ERP-waveforms for true and false recognition are expected to be equivalent. More specifically, relative to new responses of NEW items true and false recognition are expected to elicit early fronto-medial old/new ERP effects and parietal old/new ERP effects. If, however, false recognition is based only on familiarity, then it will fail to elicit a parietal old/new effect.
Section snippets
Participants
Twenty-two volunteers (13 female) between 19 and 28 years of age (mean 23 years) participated. They were students at the University of Leipzig, were right-handed and had normal or corrected-to-normal vision. They reported to be in good health and were paid 12 DM/h for their participation. None of the participants had prior experience with the task.
Experimental materials
Stimuli consisted of 300 German nouns taken from a categorical word pool. This pool was created in a categorical noun generation experiment performed
Participants
Thirty-six volunteers (25 female) participated in the experiment. They were students at the University of Leipzig and were between 20 and 32 years of age (mean: 23 years), were right handed and had normal or corrected-to-normal vision. They reported to be in good health and were paid 12 DM/h. None of the participants had any prior experience with the task.
Stimuli and procedure
We used the same word list as in Experiment 1. In the study phase, Category Group participants assigned words to a specific category, while
General discussion
Taken together, the results from the group analysis of Experiment 1 and the results from Experiment 2 support the view that participants differentially encoded categorical relations of the studied words. When participants had high false recognition rates (Experiment 1), suggesting the use of categorical relations, or when participants directly focused on categorical relations (Experiment 2; Category Group) ERP effects for true and false recognition were similar (see results summary in Table 6).
Acknowledgements
This work was supported by the Leibniz-Prize of the German Research Foundation awarded to Angela D. Friederici, Director at the Max-Planck-Institute of Cognitive Neuroscience. We wish to thank Ina Koch, Arlett Schueller, and Kerstin Weber for their valuable assistance in data collection.
References (72)
- et al.
Electrophysiological evidence for dissociable processes contributing to recollection
Acta Psychol.
(1998) - et al.
Functional-anatomic study of episodic retrieval using fMRI: I retrieval effort vs. retrieval success
Neuroimage
(1998) The electrophysiology of incidental and intentional retrieval: ERP old/new effects in lexical decision and recognition memory
Neuropsychologia
(1999)- et al.
False recognition after a right frontal lobe infarction: memory for general and specific information
Neuropsychologia
(1997) - et al.
Regional modulation of high resolution evoked potentials during verbal and non-verbal matching tasks
Electroencephalogr. Clin. Neurophysiol.
(1995) - et al.
High resolution evoked potential imaging of the cortical dynamics of human working memory
Electroencephalogr. Clin. Neurophysiol.
(1996) - et al.
Cerebral location of international 10–20 system electrode placement
Electroencephalogr. Clin. Neurophysiol.
(1987) - et al.
A spatio-temporal analysis of recognition-related event-related brain potentials
Int. J. Psychophysiol.
(1998) The persistence of false memories in list recall
J. Memory Language
(1996)On the modularity of recognition memory for object form and spatial location: a topographic ERP analysis
Neuropsychologia
(1998)