Search

Several measurement models have been proposed for data from the continuous-reproduction paradigm for studying visual working memory: The original mixture model (Zhang & Luck, 2008) and its extension (Bays, Catalao, & Husain, 2009); the interference measurement model (Oberauer, Stoneking, Wabersich, & Lin, 2017), and the target confusability competition model (Schurgin, Wixted, & Brady, 2020). This article describes a space of possible measurement models in which all existing models can be placed. The space is defined by three dimensions: (1) The choice of a activation function (von-Mises or Laplace), the choice of a response-selection function (variants of Luce’s choice rule or of signal detection theory), and whether or not memory precision is assumed to be a constant over manipulations affecting memory. A factorial combination of these three variables generates all possible models in the model space. Fitting all models to eight data sets revealed a new model as empirically most adequate, which combines a von-Mises activation function with a signal-detection response-selection rule. The precision parameter can be treated as a constant across many experimental manipulations, though it might vary with manipulations not yet explored. All modelling code and the raw data modelled are available on the OSF: osf.io/zwprv

Previous research assumes that executive functions such as inhibition, shifting and updating explain individual differences in cognitive abilities. Of these three executive functions, updating was previously found to relate most strongly to fluid intelligence. However, this relationship could be a methodological artifact: Measures of inhibition and shifting usually isolate the contribution of this executive function to performance by contrasting conditions with high and low demands on these processes, whereas updating is measured by overall accuracy in working memory tasks involving updating. This updating measure conflates updating-specific individual differences (e.g., removal of outdated information) with variance in working memory maintenance. Re-analyzing data (N = 111) from von Bastian et al. (2016), we separated updating-specific variance from working memory maintenance variance. Updating contributed only 15% to individual differences in performance in the updating tasks, and it correlated neither with fluid intelligence nor with independent working memory measures reflecting storage and processing or relational integration. In contrast, the working memory maintenance component of the updating task correlated with both abilities. These findings challenge the view that updating contributes to variance in higher cognitive abilities.

We show that the intention to remember information substantially improves recall of that information when maintained in working memory (WM), whereas it does not have the same effect on maintenance in episodic long-term memory (LTM). In eight experiments, participants processed lists of words according to a semantic orienting task in three instruction conditions: to remember, to forget, or an incidental-memory baseline with no instruction. The first two experiments showed better memory for intentional remembering compared to incidental memory for a typical test of WM, but not for a typical test of episodic LTM. The subsequent six experiments determined which of three variables distinguishing typical WM and episodic-LTM tests-delay of test, list length, and proactive interference-are responsible for this difference. The intention to remember improved free-recall performance for short lists (close to the capacity limit of WM), most strongly when tested immediately, and only in the presence of proactive interference. This result supports a functional distinction between WM and episodic LTM: Whereas episodic LTM keeps a nonselective record of experiences for future use, WM holds selectively only the information relevant for the current goal. In addition, we found a beneficial effect of intentional remembering on memory for the list position of words, which was obtained for longer lists and regardless of the delay of testing, probably reflecting episodic LTM. The instruction to forget had no effect compared to the incidental baseline, questioning the assumption of a targeted forgetting process on memory. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

This series of Experiments aims to investigate the mechanisms underlying learning from repetition in the Hebb paradigm. In 5 Experiments, we will test how the salience of different repeated list structures can facilitate learning of (complex) partially repeated patterns. The detailed preregistration is included in the file "preregistration_exp1-5.pdf" which is registered together with this registration.

One of the best-known demonstrations of long-term learning through repetition is the Hebb effect: Immediate recall of a memory list repeated amidst non-repeated lists improves steadily with repetitions. However, previous studies often failed to observe this effect for visuo-spatial arrays. Souza and Oberauer (2022) showed that the strongest determinant for producing learning was the difficulty of the test: Learning was consistently observed when participants recalled all items of a visuo-spatial array (difficult test) but not if only one item was recalled, or recognition procedures were used (less difficult tests). This suggests that long-term learning was promoted by increased testing demands over the short-term. Alternatively, it is possible that lower testing demands still lead to learning but prevented the application of what was learned. In four preregistered experiments (N = 981), we ruled out this alternative explanation: Changing the type of memory test mid-way through the experiment from less demanding (i.e., single item recall or recognition) to a more demanding test (i.e., full item recall) did not reveal hidden learning, and changing it from the more demanding to a less demanding test did not conceal learning. Mixing high and low demanding tests for non-repeated arrays, however, eventually produced Hebb learning even for the less demanding testing conditions. We propose that testing affects long-term learning in two ways: Expectations of the test difficulty influence how information is encoded into memory, and retrieval consolidates this information in memory.

This is a follow-up experiment on our initial experiment, motivated by the finding of Souza & Oberauer (2020) that change detection procedures did not lead to a credible learning effect for repeatedly presented visual arrays. To investigate the influence of different testing conditions for the long-term learning of visuospatial arrays, the aim of this experiment is to directly compare a discrete-color-reproduction task with a change-detection procedure.

Novelty-gated encoding is the assumption that events are encoded more strongly into memory when they are more novel in comparison to previously encoded events. It is a core assumption of the SOB model of serial recall (Farrell & Lewandowsky, 2002). We present three experiments testing some predictions from novelty-gated encoding. Experiment 1 shows that the probability of recalling the third item in a list correctly does not depend on whether it is preceded by phonologically similar or dissimilar items. Experiment 2 shows that in lists of items from three classes (nonwords, spatial locations, and abstract drawings) the probability of recalling an item does not depend on whether it is preceded by items from the same or another class. Experiment 3 used a complex-span paradigm varying the phonological similarity of words that are read aloud as distractors in between memory items. Contrary to a prediction from novelty-gated encoding, similar distractors did not impair memory more than dissimilar distractors. The results question the assumption of novelty-gated encoding in serial recall. We discuss alternative explanations for the phenomena that this assumption has previously helped to explain. The present evidence against novelty-gated encoding might point to boundary conditions for the role of prediction error in the acquisition of memories.

Giving people more time to process information in working memory improves their performance on working memory tasks. It is often assumed that free time given after presentation of an item enables maintenance processes to counteract forgetting of this item, suggesting that time has a retroactive benefit. Two other hypotheses—short-term consolidation and temporal distinctiveness—entail a local effect of time on immediately preceding and following items. Here, we show instead a novel global and proactive benefit of time in working memory. In three serial-recall experiments ( Ns = 21, 25, and 26 young adults, respectively), we varied the position and duration of the free time within a seven-item list of consonants. Experiment 1 showed that the effect is global and not local. Experiments 2a and 2b showed that increased interitem time benefited performance only for the subsequent items, implying a proactive benefit. This finding rules out maintenance processes, short-term consolidation, and temporal distinctiveness as explanations of the free-time benefit but is consistent with the proposal of a gradually recovering encoding resource.

Working memory (short WM) is the system responsible for holding mental representations temporarily available for later use in thought or action (Oberauer et al., 2018). Because WM has a severely limited capacity, WM content needs to be constantly updated to focus on current, relevant information. To enable seamless updating, there should be a mechanism that removes outdated information from WM. Recent studies show that under some conditions irrelevant content can be removed from WM with remarkable efficiency (e.g., Oberauer, 2018; Williams & Woodman, 2012): Oberauer (2018) had participants learn a set of six words (set-size 6 baseline) or three words (set-size 3 baseline) for a subsequent recognition test. In the recognition test, one of the previously studied words was presented in a randomly selected frame. The word could either match the word originally presented in that frame (50% of all trials) or not (mismatch, 50% of all trials). In a third condition (set size 6-3), six words were presented, but a random subset of three words was followed directly by a forget cue thereby reducing the need to remember the initial set of six memory items by three. In the set-size 6-3 condition, participants performed just as well on the later WM test of the cued to-be-remembered (TBR) items as in the set-size 3 baseline. Oberauer (2018) concluded that thus the memory items had been removed instantaneously and completely from WM. In a recent experiment we investigated what happens with the mental representations of stimuli in WM when they are cued to be forgotten. In particular, we found evidence that people are able to selectively remove irrelevant information from WM by combining the DF paradigm used in long-term memory studies (e.g., Bjork, 1970) with the item-removal paradigm in WM (Oberauer, 2018). The present study is an extension of this work: Here, we will test whether holding an item in WM longer makes it harder to subsequently remove it as it has been observed for DF in episodic memory. To gauge memory for the TBF words directly, on three trials we will probe one of the TBF words in one of the TBF frames. We will refer to this surprise test as a TBF test. We will vary the time between the offset of each word and the onset of the forget or remember cue (stimulus-cue-interval, SCI) over three levels: 1s, 2s, or 3s. Thereby we aim to investigate the boundary conditions of the selective information removal from WM. We expect that with increasing SCI, the TBF word is removed less effectively from WM, as measured by both indicators of removal: improved memory for TBR words and inability to remember TBF words. Possibly, when the time between the presentation of a word and the onset of the cue indicating the relevance of a word increases, traces of that particular word may be stronger encoded into episodic LTM.

Two Models of Memory for Serial Order Applied to Continuous Visual Stimuli

Working memory (short WM) is the system responsible for holding mental representations temporarily available for later use in thought or action (Oberauer et al., 2018). Because WM has a severely limited capacity, WM content needs to be constantly updated to focus on current, relevant information. To enable seamless updating, there should be a mechanism that removes outdated information from WM. Recent studies show that under some conditions irrelevant content can be removed from WM with remarkable efficiency (e.g., Oberauer, 2018; Williams & Woodman, 2012): Oberauer (2018) had participants learn a set of six words (set-size 6 baseline) or three words (set-size 3 baseline) for a subsequent recognition test. In the recognition test, one of the previously studied words was presented in a randomly selected frame. The word could either match the word originally presented in that frame (50% of all trials) or not (mismatch, 50% of all trials). In a third condition (set size 6-3), six words were presented, but a random subset of three words was followed directly by a forget cue thereby reducing the need to remember the initial set of six memory items by three. In the set-size 6-3 condition, participants performed just as well on the later WM test of the cued to-be-remembered (TBR) items as in the set-size 3 baseline. Oberauer (2018) concluded that thus the memory items had been removed instantaneously and completely from WM. While studies of directed forgetting (DF) in long-term memory and WM use similar methods to induce forgetting (i.e., remember/forget cues in the list-method or item-method of DF), they test for the effectiveness of forgetting in different ways: WM studies assess it only indirectly by showing that access to TBR items (or encoding of new content) is facilitated (see also Lewis-Peacock et al., 2018). Episodic-memory studies on the other hand usually gauge the availability of to-be-forgotten (TBF) information directly through a surprise test. The fact that the retrieval performance for TBF items has not been directly assessed – a notable exception to this is a study on visual WM (Williams et al., 2013; yet, only using two TBR objects) – is particularly striking as testing the TBF items in a surprise test is the basis for DF paradigms in long-term memory. The proposed project aims to shed new light on what happens with the mental representations of stimuli in WM when they are cued to be forgotten, by testing their availability directly. In particular, we will test to what extent people selectively remove irrelevant information from WM by combining the DF paradigm used in long-term memory studies (e.g., Bjork, 1970) with the item-removal paradigm in WM (Oberauer, 2018). To gauge memory for the TBF words directly, on a small subset of trials we will probe one of the TBF words in one of the TBF frames. We will refer to this surprise test as a TBF test. Doing this raises a methodological challenge: Once participants are tested for items that they were instructed to forget, they may start to ignore the forget cues (see Williams & Woodman, 2012). Prior research (Muter, 1980; Williams et al., 2013) showed that TBF items are very effectively forgotten when tested on only 1-2% of trials. Work by Dagry and Barrouillet (2017) suggests that even material tested on >20% of trials is effectively forgotten. To assess how often we can test TBF items without undermining the forget instruction, we will vary the frequency with which the TBF items are tested (5%, 10%, or 20% of all trials) between participants and assess the stability of DF effects. In sum, this experiment will allow us to directly assess whether TBF items are no longer available in WM and can thus not be retrieved in recognition tests (i.e., accuracy for TBF items should be close to chance).

There is a strong relationship between fluid intelligence and working memory capacity (WMC). Yet, the cognitive mechanisms underlying this relationship remain elusive. The capacity hypothesis states that this relationship is due to limitations in the amount of information that can be stored and held active in working memory. Previous research aimed at testing the capacity hypothesis assumed that it implies stronger relationships of intelligence test performance with WMC for test items with higher capacity demands. The present article addresses this assumption through simulations of three theoretical models implementing the capacity hypothesis while systematically varying different psychometric variables. The results show that almost any relation between the capacity demands of items and their correlation with WMC can be obtained. Therefore, the assumption made by previous studies does not hold: The capacity hypothesis does not imply stronger correlations of WMC and intelligence test items with higher capacity demands. Items varying in capacity demands cannot be used to test the causality of WMC (or any other latent variable) for fluid intelligence.

When encoding task-relevant information in working memory (WM), we can use prior knowledge to facilitate task performance. For instance, when memorizing a phone number, we can benefit from recognizing some parts as known chunks (e.g., 911) and focus on memorizing the novel parts. Prior knowledge from long-term memory (LTM), however, can also proactively interfere with WM contents. Here, we show that WM selectively recruits information from LTM only when it is helpful, not when it would interfere. We used variants of the Hebb paradigm in which WM is tested through immediate serial recall of lists. Some lists were repeated frequently across trials, so they were acquired in LTM, as reflected in increasing serial-recall performance across repetitions. We compared interference conditions in which that LTM knowledge could interfere with holding another list in WM to a neutral condition in which that knowledge could be neither beneficial nor harmful. In Experiments 1-3, lists in the interference conditions shared their items with the learned lists but not their order. We observed no proactive interference. In Experiments 4 and 5, the interference lists' first three items overlapped exactly with the learned lists, and only the remaining items had a new order. This made LTM knowledge partially beneficial and partially harmful. Participants could use LTM flexibly to improve performance for the first part of the list without experiencing interference on the second half. LTM-mediated learning of the first part even boosted memory for the unknown second part. We conclude that there is a flexible gate controlling the flow of information from LTM and WM so that LTM knowledge is recruited only when helpful. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

We introduce a framework for simple measurement models for working memory, and apply it to complex-span and memory-updating tasks. Memory Measurement Models (M³) use the frequency distribution across response categories to measure continuous memory strength along two dimensions: Memory for individual elements, potentially relying on persistent activation of unified representations, and memory for relations, relying on temporary bindings. Experiment 1 provides evidence for the validity of the parameters measuring these two dimensions of strength. The effects of experimental manipulations on these two dimensions can be captured by additional model parameters that reflect hypothetical processes affecting memory. Across five further experiments we illustrate how M³ can be used to measure three such processes: The continued strengthening of memory representations during the retention interval (extended encoding), the dampening of encoding of irrelevant information (filtering), and the removal of irrelevant information from memory. In one experiment we compare young and old adults on complex-span tasks and working-memory updating. In both paradigms, old adults showed impaired memory for relations but no impairment in memory for individual elements. There was partial evidence for age differences in extended encoding and removal; there were no age differences in filtering. We suggest that M³ offer a computationally efficient approach to identifying memory processes. All data and model codes are publicly available on the Open Science Framework: osf.io/vkhmu

A common assumption in theories of working memory is that a maintenance process - broadly referred to as rehearsal - is involved in keeping novel information available. This review evaluates the effectiveness of three forms of rehearsal: articulatory rehearsal, attention-based refreshing, and elaborative rehearsal. Evidence for the effectiveness of these strategies is surprisingly weak. Experimental manipulations of articulatory rehearsal have yielded working memory benefits in children, but not in adults; experimentally induced refreshing prioritizes the refreshed information, but yields little benefit compared to a baseline without induced refreshing; and elaborative rehearsal improves episodic long-term memory but has little effect on working memory. Thus, although adults spontaneously use some of these strategies, rehearsal might not play a causal role in keeping information in working memory.

There is a recent surge of interest in maintenance processes in working memory, such as articulatory rehearsal, elaboration, and attentional refreshing. Yet, we know little about the central attentional demand of these processes. It has been assumed that articulatory rehearsal does not require central attention at all (Vergauwe, Camos, & Barrouillet, 2014), being in essence a cost-free strategy. In contrast, elaboration and attentional refreshing are assumed to incur large and continuous costs on central attention. We tested these assumptions in three experiments in which participants were presented with a varying number of words to rehearse. Participants were instructed to rehearse the words aloud, or to elaborate them by creating interactive images. Attentional refreshing was examined in a condition in which words were to be maintained during articulatory suppression. During retention participants carried out a series of choice reaction tasks, which were used to measure central attentional demands of the maintenance strategies. Articulatory rehearsal had costs on processing RTs that lasted for 10 s. Maintenance of words during articulatory suppression did not yield persistent costs on central attention, implying that participants did not continuously refresh the words. Finally, the current results cast doubt on the idea that elaboration requires central attention for an extended period of time. All experimental scripts and data sets reported here are available online ( https://osf.io/69p8j/ ).

In tests of working memory with verbal or spatial materials, repeating the same memory sets across trials leads to improved memory performance. This well-established “Hebb repetition effect” could not be shown for visual materials in previous research. The absence of the Hebb effect can be explained in two ways: Either persons fail to acquire a long-term memory representation of the repeated memory sets, or they acquire such long-term memory representations, but fail to use them during the working memory task. In two experiments (N1 = 18 and N2 = 30), we aimed to decide between these two possibilities by manipulating the long-term memory knowledge of some of the memory sets used in a change-detection task. Before the change-detection test, participants learned three arrays of colors to criterion. The subsequent change-detection test contained both previously learned and new color arrays. Change detection performance was better on previously learned compared with new arrays, showing that long-term memory is used in change detection.

We respond to the comments of Logie and Vandierendonck to our article proposing benchmark findings for evaluating theories and models of short-term and working memory. The response focuses on the two main points of criticism: (a) Logie and Vandierendonck argue that the scope of the set of benchmarks is too narrow. We explain why findings on how working memory is used in complex cognition, findings on executive functions, and findings from neuropsychological case studies are currently not included in the benchmarks, and why findings with visual and spatial materials are less prevalent among them. (b) The critics question the usefulness of the benchmarks and their ratings for advancing theory development. We explain why selecting and rating benchmarks is important and justifiable, and acknowledge that the present selection and rating decisions are in need of continuous updating. The usefulness of the benchmarks of all ratings is also enhanced by our concomitant online posting of data for many of these benchmarks. (PsycINFO Database Record