Modulation of the word frequency effect in recognition memory after an unrelated lexical decision task

Abstract

The natural language frequency of a word is known to influence the ability to perform recognition judgments based on either semantic or episodic memory, an effect commonly known as the word frequency effect (WFE). For episodic recognition specifically, the WFE presents a mirrored pattern with higher hit rates and lower false alarm rates for low frequency words compared to high frequency words. Interestingly, the use of certain study tasks such as judgements of concreteness has been shown to reduce or even abolish the hit rate advantage of low frequency words. In the present study we tested the hypothesis that prior exposure to an unrelated lexical decision task leads to a modulation of the WFE in a subsequent episodic recognition test. Results showed an increase in hit rates and false alarm rates for high frequency words and a decrease in false alarm rates for low-frequency words. Analysis of behaviour with the linear ballistic accumulator model furthermore revealed that this effect was driven by changes in the rate of evidence accumulation of the decision process and in the response threshold. We discuss the relevance of these findings for current theories of memory.

The Discovery and Interpretation of Evidence Accumulation Stages

“…Henceforth we will refer to these durations as the decision times (DT). Because previous work relates stage 4 to a memory retrieval process (Anderson et al 2016(Anderson et al , 2018Borst et al 2016;Borst and Anderson 2015;Portoles et al 2018), it is appropriate to model the retrieval process as a noisy accumulation (e.g., Criss 2010;Neville et al 2019;Ratcliff 1978). We chose to apply the shifted Wald Model (Anders et al 2016) to model the decision times.…”

“…For example, differences in the average rate of evidence accumulation are often interpreted as differences in the difficulty to extract information from a stimulus (e.g., Donkin and Van Maanen 2014) or from memory (e.g., Wagenmakers et al 2008). Differences in the distance toward the threshold are often thought to reflect response caution (e.g., Van Maanen et al 2011) or the prior probability of the response (that is, bias, e.g., Mulder et al 2012;Neville et al 2019).…”

“…Evidence is accumulated until a certain threshold is reached, and people are confident that the stimulus represents something they have observed before or not (Gillund and Shiffrin 1984;Raaijmakers and Shiffrin 1981). Indeed, computational models that incorporate the evidence accumulation principle accurately describe the data of recognition memory tasks, such as old/new recognition judgments (Neville et al 2019;Ratcliff 1978).…”

“…Among the tasks analyzed with HsMM-MVPA were an associative recognition task and two working memory tasks. Although the structure of these tasks resembles the recognition memory tasks that were previously successfully analyzed using evidence accumulation models (Neville et al 2019;Ratcliff 1978), Anderson and colleagues found evidence for a sequence of at least five processing stages. This is an indication that the typical three-stage assumption of evidence accumulation models might not hold, even in such relatively simple tasks.…”

Analysis of Skim Reading on Desktop versus Mobile Screen

“…Therefore, for measuring sensitivity (d') and bias (C) factors such as hit rate, false alarm, miss rate, and the correct rejections were computed. A value of hit rate is obtained when the participant marked the sentences as "True" which were present in the text whereas the sentences which were in the text, but the participant marked them as "False" gives the false alarm rate [53]. Participants checked the sentences as "True" when the sentences were not in the text they just read.…”