A partial resolution of the paradox of interference: The role of integrating knowledge

It has been noted that models of memory that posit retrieval interference imply that the more one knows about a topic, the harder it is to retrieve any one of these facts. Smith, Adams, and Schorr (Cognitive Psychology, 1978, 10, 438–464) regard this to be a paradox and postulate that people use world knowledge to integrate various facts about a concept and thereby avoid interference. Exploring this issue further in two experiments we discovered that integration of facts alleviates interference only when a person can perform his memory task by simply making a consistency judgment and can avoid the need to retrieve a specific fact. When foils force subjects to retrieve the specific assertion, the interference occurs among integrated facts as among unrelated facts. It appears that, when possible, subjects will judge whether they have seen a fact simply by judging if it is related to (consistent with) a theme they have studied. In other words, people judge themes rather than facts. Consistent with this interpretation, we found interference among themes; that is, the more themes were associated with a concept, the greater the interference.     

Focused memory search in fact retrieval

Several studies of fact retrieval have shown that the more facts a person learns about a concept, the longer it takes him or her to retrieve any of these facts. This result has been interpreted to mean that retrieval of a fact about a concept involves a search of all facts stored in memory with that concept. In the present study, it is suggested that retrieval involves not an unfocused search of all facts stored with a concept, but rather a focused memory search that examines relevant stored facts and ignores irrelevant information. This argument is supported by three experiments in which subjects first learned simple facts (e.g., “The banker likes horses”) and then made speeded true-false decisions for test probes le.g., “The banker likes elephants”). Specifically, results suggest that facts stored with a concept may be organized into subsets. For example, a person’s knowledge about Richard Nixon might be organized into subsets concerning Nixon’s resignation, his trips to China, his family, and so on. The data further suggest that a person attempting to retrieve a fact about a concept (e.g., the name of Nixon’s wife) may simply decide which subset is most likely to contain the desired fact (e.g., the subset concerning Nixon’s family) and search that subset. If the sought-for fact is found in this subset, the search process terminates. If, however, the desired information is not located, other subsets of facts may be searched before the retrieval attempt is given up. The notion that memory search focuses on relevant stored facts and ignores irrelevant information may help to explain why experts (i.e., people who know a large number of facts about a topic) do not experience great difficulty in retrieving facts in their areas of expertise.     

Situation models,propositions, and the fan effect

According to theories of language comprehension, people can construct multiple levels of representation: the surface form, the propositional text base, and the situation model. In this study, I looked at how the referential nature of memory probes affects the experience of retrieval interference. All the subjects memorized sentences about objects in locations (e.g., “The potted palm is in the hotel”). When memory probes were sentences and, therefore, referential and most closely associated with the situation model level, no interference was observed during retrieval for information that could be integrated into a common situation model. In contrast, interference was observed in such cases when the memory probes were concept pairs (POTTED PALM-HOTEL) and hence not directly referential. This is interpreted to mean that nonreferential memory probes involve surface form and text base representations more than do referential sentence probes.     

Eye movements do not reflect retrieval processes: limits of the eye-mind hypothesis

Abstract— This research investigated whether eye movements are informative about retrieval processes. Participants learned facts about persons and locations, and the number of facts (fan) learned about each person and location was manipulated. During a subsequent recognition test, participants made more gazes to high-fan facts than to low-fan facts, and gazes to high-fan facts had a longer duration than gazes to low-fan facts. However, there was no relation between the order in which items were fixated and the relative effect of person or location fan. The effect of person and location fan on gaze duration also did not differ with whether it was the person or location being fixated. A model assuming that the process of retrieval is independent of eye movements was successfully fit to the data on the distribution of gaze durations. According to this model, the effect of fan on number of gazes and gaze duration is an artifact of the longer retrieval times for high-fan facts.     

Differential fan effect and attentional focus

As people study more facts about a concept, it takes longer to retrieve a particular fact about that concept. This fan effect (Anderson, 1974) has been attributed to competition among associations to a concept. Alternatively, the mental-model theory (Radvansky & Zacks, 1991) suggests that the fan effect disappears when the related concepts are organized into a single mental model. In the present study, attentional focus was manipulated to affect the mental model to be constructed. One group of participants focused on the person dimension of personlocation pairs, whereas the other group focused on the location dimension. The result showed that the fan effect with the focused dimension was greater than the fan effect with the nonfocused dimension, which is contrary to the mental-model theory. The number of associations with a concept is indeed crucial during retrieval, and the importance of the information seems to be accentuated with attentional focus.     

Storage and retrieval processes in sentence memory

Two experiments investigated how much time Ss take to verify a sentence with respect to a set of subject-verb-object (SVO) propositions they had learned. Three triplets of SVOs provided differing degrees of equivocation, including (1) unique SVOs, (2) a V with 3 SOs, and (3) an SV with 3 Os. A general class of models was proposed regarding the representation of prepositional trees in memory and an algorithm by which a probe tree might be verified against a set of trees in memory. In Expt I, verification times were least for Condition (1), slower and equal for Conditions (2) and (3). In Expt II, a fourth condition, an SO with 3 Vs, produced verification times as slow as those in Condition (3). Of several models considered, the closest-fitting one assumed parallel search processes proceeding simultaneously from S, V, and O positions of the probe, scanning through memory trees structered according to the HAM (Human Associative Memory) theory of Anderson and Bower.     

Influence of emotional state on irrelevant thoughts

The fan effect paradigm was used to investigate the influence of emotional mood state on longterm memory retrieval (Anderson, 1983). Subjects learned target facts embedded in unrelated sentences to a specified criterion and were then given a happy, sad, or neutral mood induction. Mean response times (RTs) and error rates were analyzed in a speeded recognition test in which subjects distinguished between the learned facts and foil facts (foil facts were constructed by recombining the same concepts). A follow-up lexical decision task indicated that mean RT was positively correlated with an increase in the weighted proportion of irrelevant thoughts produced by subjects in an induced sad mood. Results suggest that irrelevant thoughts associated with the sad mood state interfered with more relevant, task-oriented, thoughts and support the notion that sad mood is related to a failure to inhibit irrelevant information.     

Affective integration speeds trait processing

We investigated whether affective integration increases the speed of processing of personality trait knowledge. The fan effect was compared between cases where trait knowledge is stored with the affective value and cases where it is not stored with the affective value. 18 college students first memorized a set of traits about fictitious individuals and then made recognition judgments. In the 2 x 2 factorial repeated-measures design, the number of traits learned about a fictitious individual and whether those traits were integrated by a shared affective value were manipulated. The significant interaction showed that knowledge of personality trait with affective integration was processed quickly even if the particular person's memory had rich connections with traits.     

Individuation and reference in memory: Proper names and definite descriptions

A theory is presented which assumes that individuals are represented by unique nodes in memory. To test the theory, simple facts were predicated of an individual person. Some facts referred to him by proper name, and other facts by his profession. In a before condition, subjects learned that the profession and name referred to the same individual before learning the facts, while in an after condition, they learned the identity after learning the facts. Subsequent to learning the facts and identities, subjects verified sentences based on what they had learned. Verification latencies indicated that in the before condition, one memory node was created to represent the individual, but two nodes were set up in the after condition. Assymmetries between proper names and professions indicate that the two types of referring expressions are treated differently in long-term memory.     

Spatial directions and situation model organization

Do spatial directions, such as “to the right,” influence the integration and segregation of information into situation models? According to a single-framework hypothesis, spatial location serves as an event framework, and spatial directions serve as relational information within that framework but do not establish separate sublocation frameworks. Alternatively, according to a fragmented-framework hypothesis, spatial directions lead the larger framework to be broken down such that each direction is treated as a separate sublocation, thereby producing retrieval interference. In three experiments, people memorized sentences about objects in locations. The results support the fragmented-framework hypothesis. Control conditions ruled out explanations based on the ease of memorization, retrieval demands, or sentence complexity.     

Recognition and retrieval of proper names: Age differences in the fan effect

Abstract

A spreading activation model was applied, to explain. some aspects of retrieval failure for proper names. The fact that retrieval failures occur more frequently for well-known names than for unfamiliar names, and the increased incidence of name blocking in the elderly, are both interpreted in terms of the fan effect. Fanning is created when multiple far are known about a single person and results in slower and more error-prone retrieval, but the fan effect can be counteracted if the facts form an integrated set. Three groups of subjects—young, middle-aged, and elderly—learned sets of name-attribute statements in three conditions. In the no-fan condition, each name was uniquely paired with a single attribute; in the fan condition, each name was linked with three different attributes; in the crossed-fan condition, different names shared some of the same attributes. Verification RTs and cued recall responses showed that all subjects found the crossed-fan condition most difficult, but, when name-attribute links were not crossed, young and middle-aged subjects were able to counteract the fan effect by integrating the facts into a unified representation. Elderly subjects were less able to integrate effectively and were more susceptible to the fan effect.     

Interference with real world knowledge

Two experiments are described in which subjects studied made-up, fantasy facts about well-known persons and then were asked to verify actual facts about these persons. Reaction time to the actual facts was longer the more fantasy propositions studied about a person. Reaction time was also longer when the verification test involved a mixture of actual and fantasy facts rather than just actual facts. A mathematical version of the ACT model (Anderson, 1976) was fit to the data. It provides a satisfactory fit, better than an alternate model. However, some of the parameter values estimated for the ACT model seemed unreasonable.     

The organization of information retrieved from situation models

The organization of information when it is retrieved from situation models was explored using recognition and recall tests. A fan effect paradigm was used to assess organization in recognition and a clustering measure was used in recall. People memorized either object-location facts (e.g., “The pay phone is in the city hall”) or person/small location facts (e.g., “The banker is on the witness stand”). For object-location facts, a location-based organization was observed in both recall and recognition. However, for person/small location facts, a location-based organization was observed for recall, and a person-based organization was observed for recognition. Thus, the observed organization was flexible depending on the retrieval task.     

The effects of memory set size and information structure on learning and retention

Two experiments examined the effects of memory set size and information structure on learning and retention. Participants learned 48 (small set) or 144 (large set) facts about individuals, and were tested over 48 facts. The test facts included either 4 facts about 12 individuals (12-person condition) or 12 facts about 4 individuals (4-person condition). During learning, there was an advantage for the small-set group in the 4-person condition, but a disadvantage in the 12-person condition. During testing, there was an advantage for the 4-person condition relative to the 12-person condition for the small-set group, even when the conditions were equated in terms of name exposure. The results support a mental model account of memory representation and retrieval.     

Memory retrieval and suppression: the inhibition of situation models

When people retrieve newly learned facts on a recognition test, they are often increasingly slowed by the number of other newly learned facts that have a concept in common with the probed fact. This is called the fan effect. Assuming that people are using situation models of the learned information, the author considers whether the inhibition of competing representations is one of the processes involved in the fan effect. Evidence was found for negative priming of related but irrelevant situation models, thus supporting the idea that the inhibition of highly related memory traces is used in long-term memory retrieval. As such, this is a form of retrieval-based inhibition.     

Retrieval of propositional information from long-term memory

Three experiments are reported in which subjects learn propositions like A hippie is in the park. The experiments manipulate the number of such propositions involving a particular person (e.g., hippie) or a particular location (e.g., park). After learning the material, subjects are asked to judge whether particular probe propositions are from the study set. Times to make these judgments about probe propositions increase with the number of study propositions involving the person or location used in the probe proposition. A model is presented which assumes a subject simultaneously accesses memory from all concepts in a probe proposition and serially searches through all study propositions involving each concept. Search of memory terminates as soon as one search process from a concept finds the probe proposition or exhausts the study propositions attached to that concept.     

Event models and the fan effect

The current study explored the persistence of event model organizations and how this influences the experience of interference during retrieval. People in this study memorized lists of sentences about objects in locations, such as “The potted palm is in the hotel.” Previous work has shown that such information can either be stored in separate event models, thereby producing retrieval interference, or integrated into common event models, thereby eliminating retrieval interference. Unlike prior studies, the current work explored the impact of forgetting up to 2 weeks later on this pattern of performance. We explored three possible outcomes across the various retention intervals. First, consistent with research showing that longer delays reduce proactive and retroactive interference, any retrieval interference effects of competing event models could be reduced over time. Second, the binding of information into events models may weaken over time, causing interference effects to emerge when they had previously been absent. Third, and finally, the organization of information into event models could remain stable over long periods of time. The results reported here are most consistent with the last outcome. While there were some minor variations across the various retention intervals, the basic pattern of event model organization remained preserved over the two-week retention period.     

Forgetting our facts: the role of inhibitory processes in the loss of propositional knowledge

Seven experiments are reported that show that retrieving facts from long-term memory is accomplished, in part, by inhibitory processes that suppress interfering facts. When asked to repeatedly retrieve a recently learned proposition (e.g., recalling The actor is looking at the tulip, given cues such as Actor looking t__), subjects experienced a recall deficit for related facts (e.g., The actor is looking at the violin) on a recall test administered 15 min later. Importantly, this retrieval-induced forgetting was shown to generalize to other facts in which the inhibited concepts took part (e.g., The teacher is lifting the violin), replicating a finding observed by M. C. Anderson and B. A. Spellman (1995) with categorical stimuli. These findings suggest a critical role for suppression in models of propositional retrieval and implicate the mere retrieval of what we know as a source of forgetting of factual knowledge.     

Memory-load interference in syntactic processing

Participants remembered a short set of words while reading syntactically complex sentences (object–extracted clefts) and syntactically simpler sentences (subject–extracted clefts) in a memory–load study. The study also manipulated whether the words in the set and the words in the sentence were of matched or unmatched types (common nouns vs. proper names). Performance in sentence comprehension was worse for complex sentences than for simpler sentences, and this effect was greater when the type of words in the memory load matched the type of words in the sentence. These results indicate that syntactic processing is not modular, instead suggesting that it relies on working memory resources that are used for other nonsyntactic processes. Further, the results indicate that similarity–based interference is an important constraint on information processing that can be overcome to some degree during language comprehension by using the coherence of language to construct integrated representations of meaning.