Validating information during reading: the effect of recency

Please download to get full document.

View again

of 17
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Similar Documents
Information Report
Category:

Internet & Web

Published:

Views: 0 | Pages: 17

Extension: PDF | Download: 0

Share
Description
Previous studies demonstrated that outdated information may be reactivated and disrupt subsequent processing of newly encoded information. However, previous studies focused on the impact of outdated information that had been backgrounded in memory.
Tags
Transcript
  Validating information during reading:the effect of recency Sabine Guéraud University of Paris 8, Saint-Denis, France Erinn K. Walsh University of New Hampshire, Durham, NH, USA Anne E. Cook University of Utah, Salt Lake City, UT, USA Edward J. O ’ Brien University of New Hampshire, Durham, NH, USA Previous studies demonstrated that outdated information may be reactivated and dis-rupt subsequent processing of newly encoded information. However, previous studiesfocused on the impact of outdated information that had been backgrounded in memory.The present experiments examined the immediate in 󿬂 uence of outdated information;backgrounding information was reduced so that both outdated and current protagonist characteristics were active in memory when readers encountered content that wasinconsistent with the outdated information. In Experiments 1 and 2, the order of intro-duction of outdated and current information was varied; across the two experiments,the elaboration of the current information was also manipulated. When both current and outdated information are still active in memory, only the most recent informationin 󿬂 uenced comprehension; this was true regardless of whether it was elaborated. Theseresults inform the conditions under which outdated information in 󿬂 uences comprehen-sion and are interpreted within the context of the RI-Val model of comprehension. Highlights What is already known about this topic •  Previous studies have demonstrated that whether outdated information in 󿬂 uencesthe validation process depends on how quickly readers gain access to this informa-tion (e.g., O ’ Brien, Cook, & Guéraud, 2010; O ’ Brien, Rizzella, Albrecht, &Halleran, 1998); that is, validation is mediated by factors that affect the reactivationprocess (Cook & O ’ Brien, 2014). •  When information has been backgrounded in memory (i.e., when outdatedinformation is not currently active), the order in which updated versus outdatedinformation occurred does not differentially impact the validation process(Guéraud, Harmon, & Peracchi, 2005). Copyright © 2018 UKLA. Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ,UK and 350 Main Street, Malden, MA 02148, USA  Journal of Research in Reading, ISSN 0141-0423 DOI:10.1111/1467-9817.12244Volume 00, Issue 00, 2018, pp 1  –  17   What this paper adds •  Previous research has focused on how outdated information that had beenbackgrounded in memory was subsequently reactivated and in 󿬂 uenced valida-tion. However, factors that directly in 󿬂 uence the validation process remainunknown. •  In this set of studies, we explored whether memory factors such as elaboration andrecency of presentation of information directly in 󿬂 uence validation when the infor-mation in question is already active in memory, that is, when outdated informationco-occurred in memory with updated, current information. •  Resultsdemonstratedthatwhen current andoutdated information co-occur inmem-ory,onlythemostrecentinformationimpactsthevalidationprocess(Experiment1);elaborating on the current information did not override this effect (Experiment 2).  Implications for theory •  This research adds to the overall understanding of how the contents of activememory in 󿬂 uence the validation process. A basic component of comprehension involves the development of a coherent represen-tation in memory that re 󿬂 ects the intended meaning of a text. This involves the contin-ual integration of incoming information with the emerging representation. Most current models of comprehension describe the development of such representations as a two-stage process: a passive activation mechanism operates during the  󿬁 rst phase of process-ing, followed by an integration phase that links activated information to the informationcurrently being processed (e.g., Kintsch, 1988, 1998; Long & Lea, 2005; Rizzella &O ’ Brien, 1996, 2002; Sanford & Garrod, 1989, 1998, 2005). Recent research, however,suggests that comprehension also involves a third stage, a validation process that matches the linkages formed by integration against the contents of active memory(Cook & O ’ Brien, 2014; Isberner & Richter, 2013, 2014; O ’ Brien & Cook, 2016a,2016b; Singer, 2006, 2013; Singer & Doering, 2014). The goal of the present studywas to expand on these recent   󿬁 ndings by investigating the factors that govern thevalidation process.The RI-Val model, recently proposed by Cook and O ’ Brien (2014; O ’ Brien & Cook,2016a, 2016b), breaks processing into three separate but overlapping stages of reactivation(  R ), integration (  I  ) and validation ( Val  ). In the  󿬁 rst stage,  R , when new information isencoded, any information in long-term memory that is suf  󿬁 ciently related is reactivatedvia a resonance-like mechanism (Myers & O ’ Brien, 1998; O ’ Brien & Myers, 1999). Theresonance process has been described as  ‘ dumb ’ , in that information is reactivated as a function of its relatedness to information in active memory, regardless of whether it is cur-rently relevant. In the second stage,  I  , reactivated information that exceeds a minimum threshold of activation is integrated with or linked to the contents of active memory. Easeof this integration process is based on the goodness of   󿬁 t between recently encoded andreactivated information (Cook, 2014). Thus, it is possible that this initial integration pro-cess could result in the connection of related but contradictory pieces of information. Link-ages formed in this stage are then subsequently validated, or matched, against the contentsof active memory in the third,  Val  , stage. Validation is based on a passive pattern-matchingprocess (e.g., Kamas & Reder, 1995; Kamas, Reder, & Ayers, 1996; Reder & Kusbit, 1991) 2 GUÉRAUD, WALSH, COOK & O ’ BRIEN Copyright © 2018 UKLA  in which the initial linkages formed by integration are compared at a featural level against the entire contents of active memory. This validation process thus allows incoming content to be compared, or evaluated, against the contents of active memory on a continuous basis.Comprehension dif  󿬁 culty at any point is a function of the degree of match that arises from the passive matching process.A critical assumption within the RI-Val model is that activation, integration and valida-tion are all passive, parallel and asynchronous processes. Once started, these three compo-nent processes operate continuously as the reader proceeds through the text, with eachprocess operating on the output of the others. Even as validation is running, activationand integration continue making new information available for the validation component.The RI-Val model ’ s assumptions have two important consequences for comprehension.First, because validation is dependent upon the output of the previous stages, it is mediatedby factors that in 󿬂 uence those stages  –  especially the  R  stage (Cook & O ’ Brien, 2014). In-formation that is highly related to incoming content is more likely to be reactivated, inte-grated and validated before information that is weakly related to incoming information.Second, the assumption that processing is continuous raises the possibility that processingeffects may be observed at any point during comprehension; that is, processing effects maybe immediate, they may affect immediate processing and spillover to subsequent informa-tion, or they may be observed only after a delay.This timing of processing effects has been an important issue in studies that focused onthe circumstances under which inactive information (i.e., information from previous partsof the text in long-term memory) becomes reactivated during reading and in 󿬂 uences com-prehension. In short, the timing of processing effects is dependent upon how quickly theinformation in question is reactivated during reading and integrated with incoming infor-mation. Most studies on this topic in discourse comprehension have used the inconsistencyparadigm developed by O ’ Brien and colleagues (e.g., Albrecht & O ’ Brien, 1993; Cook,Halleran, & O ’ Brien, 1998; O ’ Brien & Albrecht, 1992) and have used this paradigm to in-vestigate the impact of outdated information on subsequent comprehension processes(Guéraud et al., 2005; Kendeou, Smith, & O ’ Brien, 2013; O ’ Brien et al., 1998; O ’ Brienet al., 2010; O ’ Brien, Cook, & Peracchi, 2004). For example, O ’ Brien et al. (1998) usedpassages in which a critical sentence  ‘ Mary ordered a cheeseburger and fries ’  was precededeither by consistent information (e.g., Mary loves junk food), inconsistent information(e.g., Mary is a vegetarian) or inconsistent, but quali 󿬁 ed information that indicated the pro-tagonist characteristic (e.g., vegetarian) was no longer, or never had been, true (e.g., Marywas a vegetarian, but she isn ’ t anymore). In this last Quali 󿬁 ed condition, although theprotagonist characteristic was related to and inconsistent with the contents of the criticalsentence, it was no longer relevant because it was outdated. The target sentence and thepreceding description of protagonist characteristics were separated by a backgroundingsection of about six sentences length, in order to ensure that concepts presented beforethe backgrounding section were no longer active in memory (e.g., Myers, O ’ Brien,Albrecht, & Mason, 1994). Consistent with previous studies using the inconsistency para-digm, reading times on the target sentence were slower in the Inconsistent condition than inthe Consistent condition, demonstrating that when participants encountered the target sentence, they reactivated the protagonist characteristics and attempted to integrate that information and validate it against the contents of active memory; this was easy in theConsistent condition but led to processing dif  󿬁 culty in the Inconsistent condition. Moreimportantly, reading times in the Quali 󿬁 ed condition were also slower than in the Consis-tent condition, providing evidence that even though the outdated information was no VALIDATION DURING READING 3 Copyright © 2018 UKLA  longer relevant, it was still reactivated and in 󿬂 uenced subsequent integration and validationprocesses.O ’ Brien et al. (1998; see also O ’ Brien et al., 2004) argued that encoding of the target sentence triggers reactivation of the previously presented information about the protagonist via a low-level resonance mechanism. Because the resonance process is  ‘ dumb ’ , outdatedinformation was reactivated based on its relatedness to information in active memory, eventhough it was currently not relevant. In a stronger test of this hypothesis, O ’ Brien et al.(2010) used passages in which there was an irreversible change in state of a primary object in a narrative. For example, in the Consistent condition, a once-standing tree is cut down.In the Inconsistent condition, the tree is not cut down. In the Quali 󿬁 ed condition, the tree isnot initially cut down, but then, it was struck by lightning and must be cut down. The sub-sequent target sentence was consistent with the change in state (e.g., all that remained of the tree was a stump). O ’ Brien et al. (2010) found the same pattern of results as O ’ Brienet al. (1998); the outdated information in the Quali 󿬁 ed condition was reactivated and hada negative impact on processing of the target sentence. It is important to note that in bothstudies, this occurred even though separate norming experiments demonstrated that readersclearly understood the nature of the quali 󿬁 cation and updated their discourse representa-tions in memory. Readers were presented with passages that were truncated after the last line preceding the target sentence and were asked to answer a question about the inconsis-tent content (e.g., Is Mary currently a vegetarian?); readers overwhelmingly answered  ‘ no, ’ indicating that they understood that the information was outdated and had updated their memory representation of the text before they encountered the target sentence.In follow-up to the O ’ Brien et al. (1998, 2004, 2010) studies, subsequent researchers fo-cused on the conditions that governed the reactivation and in 󿬂 uence of outdated informa-tion during reading. For example, Cook et al. (2014) demonstrated that the reactivatedoutdated information can also lead to the activation of additional inappropriate informationfrom general world knowledge  –  in their case, inappropriate predictive inferences. Guéraudet al. (2005) found that when the quali 󿬁 cation was elaborated (e.g., Mary was a vegetarian,but she isn ’ t anymore. Recently, Mary had begun to eat more meat and fast food. She es-pecially loved hamburgers), the outdated information was still reactivated in memory, asindicated by naming times in a probe task, but it was no longer suf  󿬁 cient to cause process-ing dif  󿬁 culty on the target sentence. Kendeou et al. (2013) showed that adding a causal jus-ti 󿬁 cation to the quali 󿬁 cation of the inconsistent information (e.g., Mary started eating meat because she wasn ’ t getting enough iron) was suf  󿬁 cient to eliminate any measurable reacti-vation or in 󿬂 uence of the outdated information (see also Rapp & Kendeou, 2007, 2009). Inboth the Guéraud et al. and the Kendeou et al. studies, the additional content in the quali- 󿬁 cation was suf  󿬁 cient to draw activation away from the outdated information. In the  󿬁 rst case, the outdated information was reactivated, but its level of activation was not suf  󿬁 cient to impact the validation process; in the second case, causal justi 󿬁 cation resulted in no mea-surable activation of the outdated information.Guéraud et al. (2005) also investigated whether the order in which information was intro-duced had an impact on subsequent reactivation and in 󿬂 uence of the outdated information.In the O ’ Brien et al. (1998) study, the protagonist was always introduced with the criticalcharacteristic (e.g., Mary was a vegetarian) as a de 󿬁 ning part of the protagonist  ’ s pro 󿬁 le.It was only after this information was introduced that the character trait was discounted(e.g.,  … but shewasn ’ t anymore;now she loves junk food). It was thus possible that becausethe reader   󿬁 rst learns that   ‘ Mary is a vegetarian ’ , this information may be special, resultingin a detailed representation based on this information (e.g., Peracchi & O ’ Brien, 2004; 4 GUÉRAUD, WALSH, COOK & O ’ BRIEN Copyright © 2018 UKLA  Rapp, Gerrig, & Prentice, 2001; Rapp & Kendeou, 2007, 2009); reading subsequent infor-mation that simply refutes it may not be suf  󿬁 cient to override its impact later in the text. Inorder to address this concern, Guéraud et al. (2005) varied the order in which the outdatedinformation and the qualifying information were presented. In a   ‘ Quali 󿬁 ed-First  ’  condition,the qualifying information was presented  󿬁 rst, followed by the outdated characteristic (e.g.,Marywas nolongeravegetarian, butshe usedtobeone). Ina  ‘ Quali 󿬁 ed-Second ’ condition,the outdated characteristic was presented  󿬁 rst, followed by the qualifying information (e.g.,Mary had been a vegetarian, but she was no longer one). These descriptions werebackgrounded with several sentences before the critical sentence was presented. Guéraudet al. found that varying whether the quali 󿬁 cation appeared  󿬁 rst or last had no impact onreading times on the target sentence; in both cases, readers continued to experience dif  󿬁 -culty processing the target sentences in the Quali 󿬁 ed condition.The results of the aforementioned studies can be explained within the context of the RI-Val model (Cook & O ’ Brien, 2014; O ’ Brien & Cook, 2016a, 2016b). In the  󿬁 rst stage,  R ,when the target sentence is encoded, any information in long-term memory that is suf  󿬁 -ciently related is reactivated via a dumb resonance-like mechanism (Myers & O ’ Brien,1998; O ’ Brien & Myers, 1999). Because outdated information is related to the contentsof the target sentence, it is likely to be reactivated, as demonstrated by the O ’ Brien et al.(1998, 2004, 2010) studies. However, as Kendeou et al. (2013) and Guéraud et al.(2005) demonstrated, both the content and amount of information in the quali 󿬁 cationcan in 󿬂 uence this reactivation process. In the second  I   stage of the RI-Val model,reactivated information would be linked with the contents of active memory, which in-cludes the target sentence. These linkages are subsequently validated against all of activememory in the third,  Val  , stage, via a low-level pattern matching process. Cook andO ’ Brien (2014; O ’ Brien & Cook, 2016a, 2016b) recently argued that it is when readers at-tempt to validate those linkages against information in active memory that processing dif- 󿬁 culty arises. Both the outdated characteristic (e.g., vegetarian) as well as the quali 󿬁 cation(e.g., not any more) should be reactivated during the activation stage, independent of order of mention. When this information is subsequently integrated with the target sentence andvalidated, the mismatch between Mary ’ s eating habits and her food choice (e.g., cheese-burger) arises, and comprehension dif  󿬁 culty occurs.The studies cited in the preceding paragraphs focused on how outdated information that had been backgrounded in memory was subsequently reactivated and in 󿬂 uenced compre-hension. During the validation stage within the RI-Val model, new linkages are evaluatedagainst related information in active memory. In this situation, the RI-Val model assumesthat whether previously presented information will impact or dominate initial validationand lead to comprehension dif  󿬁 culty depends on how quickly that information is madeavailable in active memory; that is, validation is mediated by factors that affect the  R  stage.What is unknown, though, is what factors directly in 󿬂 uence the validation process. Oneway to address this question is to examine the impact of factors on validation when infor-mation is already active in memory; that is, the information in question was never backgrounded and therefore does not need to be reactivated from long-term memory.Given that O ’ Brien et al. (1998, 2004, 2010) demonstrated the impact of backgroundedand then reactivated outdated information on the validation process, our focus in this studywas on the in 󿬂 uence of still-active outdated information on the validation process. Thisprovides an opportunity for further investigation into the nature of the in 󿬂 uence of activememory on validation. Two experiments were designed to assess whether memory factorssuch as elaboration and recency of presentation of information directly in 󿬂 uence validation VALIDATION DURING READING 5 Copyright © 2018 UKLA
Recommended
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x