Asymmetric negative transfer effects of working memory training.

Gathercole et al. (Journal of Memory and Language, 105, 19-42, 2019) presented a cognitive routine framework for explaining the underlying mechanisms of working memory (WM) training and transfer. This framework conceptualizes training-induced changes as the acquisition of novel cognitive routines similar to learning a new skill. We further infer that WM training might not always generate positive outcomes because previously acquired routines may affect subsequent task performance in various ways. Thus, the present study aimed to demonstrate the negative effects of WM training via two experiments. We conducted Experiment 1 online using a two-phase training paradigm with only three training sessions per phase and replicated the key findings of Gathercole and Norris (in prep.) that training on a backward circle span task (a spatial task) transferred negatively to subsequent training on a backward letter span task (a verbal task). We conducted Experiment 2 using a reversed task order design corresponding to Experiment 1. The results indicated that the transfer from backward letter training to backward circle training was not negative, but rather weakly positive, suggesting that the direction of the negative transfer effect is asymmetric. The present study therefore found that a negative transfer effect can indeed occur under certain WM training designs. The presence of this asymmetric effect indicates that backward circle and backward letter tasks require different optimal routines and that the locus of negative transfer might be the acquisition process of such optimal routines. Hence, the routines already established for backward circle might hinder the development of optimal routines for backward letter, but not vice versa.

Sequence learning recodes cortical representations instead of strengthening initial ones.

We contrast two computational models of sequence learning. The associative learner posits that learning proceeds by strengthening existing association weights. Alternatively, recoding posits that learning creates new and more efficient representations of the learned sequences. Importantly, both models propose that humans act as optimal learners but capture different statistics of the stimuli in their internal model. Furthermore, these models make dissociable predictions as to how learning changes the neural representation of sequences. We tested these predictions by using fMRI to extract neural activity patterns from the dorsal visual processing stream during a sequence recall task. We observed that only the recoding account can explain the similarity of neural activity patterns, suggesting that participants recode the learned sequences using chunks. We show that associative learning can theoretically store only very limited number of overlapping sequences, such as common in ecological working memory tasks, and hence an efficient learner should recode initial sequence representations.

Letter identity and visual similarity in the processing of diacritic letters.

Are letters with a diacritic (e.g., â) recognized as a variant of the base letter (e.g., a), or as a separate letter identity? Two recent masked priming studies, one in French and one in Spanish, investigated this question, concluding that this depends on the language-specific linguistic function served by the diacritic. Experiment 1 tested this linguistic function hypothesis using Japanese kana, in which diacritics signal consonant voicing, and like French and unlike Spanish, provide lexical contrast. Contrary to the hypothesis, Japanese kana yielded the pattern of diacritic priming like Spanish. Specifically, for a target kana with a diacritic (e.g., ガ, /ga/), the kana prime without the diacritic (e.g., カ, /ka/) facilitated recognition almost as much as the identity prime (e.g., ガ-ガ = カ-ガ), whereas for a target kana without a diacritic, the kana prime with the diacritic produced less facilitation than the identity prime (e.g., カ-カ < ガ-カ). We suggest that the pattern of diacritic priming has little to do with linguistic function, and instead it stems from a general property of visual object recognition. Experiment 2 tested this hypothesis using visually similar letters of the Latin alphabet that differ in the presence/absence of a visual feature (e.g., O and Q). The same asymmetry in priming was observed. These findings are consistent with the noisy channel model of letter/word recognition (Norris & Kinoshita, Psychological Review, 119, 517-545, 2012a).

How do we perform backward serial recall?

Following Conrad (1965, Journal of Verbal Learning and Verbal Behavior, 4, 161-169) it is often assumed that backward verbal serial recall is performed by repeated forward scans through the list and then recalling the last remaining item. Direct evidence for this peel-off strategy is relatively weak, and there has to date been no examination of its potential role in the recall of spatial sequences. To examine the role of this strategy in both verbal and spatial domains, two experiments examined response output times for forward and backward recall. For spatial span, the pattern of timing was the same in both directions. For digit span, backward recall was considerably slower. This was true whether responses were made by means of manual selection on a keyboard display (Experiment 1) or were spoken (Experiment 2a). Only two of 24 participants showed signs of using a peel-off strategy in spoken backward recall. Peel-off was not a dominant strategy in backward digit recall and there was no indication that it was ever used for spatial stimuli. Most participants reported using a combination of different strategies. In Experiment 2b, four further participants were directly instructed to use a peel-off strategy. The pattern of response times for three of these individuals was similar to the two participants from Experiment 2a previously identified as using peel-off. We conclude that backward recall can be performed using many strategies, but that the peel-off is rarely used spontaneously.

Can short-term memory be trained?

Is the capacity of short-term memory fixed, or does it improve with practice? It is already known that training on complex working memory tasks is more likely to transfer to untrained tasks with similar properties, but this approach has not been extended to the more basic short-term memory system responsible for verbal serial recall. Here we investigated this with adaptive training algorithms widely applied in working memory training. Serial recall of visually presented digits was found to improve over the course of 20 training sessions, but this improvement did not extend to recall of either spoken digits or visually presented letters. In contrast, training on a nonserial visual short-term memory color change detection task did transfer to a line orientation change detection task. We suggest that training only generates substantial transfer when the unfamiliar demands of the training activities require the development of novel routines that can then be applied to untrained versions of the same paradigm (Gathercole, Dunning, Holmes, & Norris, 2019). In contrast, serial recall of digits is fully supported by the existing verbal short-term memory system and does not require the development of new routines.

The effect of processing load on loss of information from short-term memory.

We report an experiment in which we varied the nature of the articulatory suppression task being performed during a filled retention interval in serial recall. During the retention interval participants performed one of three computer-paced colour naming tasks designed to prevent subvocal rehearsal: A Stroop color-interference task with items presented at a rate of one every 750 ms, and two color-consistent control tasks at a rate of either 750 ms or 500 ms per item. Memory performance over a 12 s interval declined much more dramatically with the Stroop task and the 500 ms control task than with the 750 ms control. There was no difference between the Stroop condition and the 500 ms control. These results pose problems for models that assume that loss of information from memory is determined entirely by interference, as there are more interfering events in the control 500 ms condition than the 750 ms Stroop. They also pose problems for models relying solely on time-based decay and articulatory rehearsal because all three conditions should block rehearsal and produce equivalent performance. The results illustrate that articulatory suppression tasks are not all equivalent, and suggest that the rate of decay from short-term memory is strongly influenced by the resource demands of concurrent processing.

Phonological recoding under articulatory suppression.

We report data from an experiment in which participants performed immediate serial recall of visually presented words with or without articulatory suppression, while also performing homophone or rhyme detection. The separation between homophonous or rhyming pairs in the list was varied. According to the working memory model (Baddeley, 1986; Baddeley & Hitch, 1974), suppression should prevent articulatory recoding. Nevertheless, rhyme and homophone detection was well above chance. However, with suppression, participants showed a greater tendency to false-alarm to orthographically related foils (e.g., GIVE-FIVE). This pattern is similar to that observed in short-term memory patients.

Learning nonwords: the Hebb repetition effect as a model of word learning.

Page and Norris [(2008). Is there a common mechanism underlying word-form learning and the Hebb repetition effect? Experimental data and a modelling framework. In A. Thorn & M. P. A. Page (Eds.), Interactions between short-term and long-term memory in the verbal domain; (2009). A model linking immediate serial recall, the Hebb repetition effect and the learning of phonological word forms. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1536), 3737-3753. doi:10.1098/rstb.2009.0173] have suggested that the Hebb [(1961). Distinctive features of learning in the higher animal. In J. F. Delafresnaye (Ed.), Brain mechanisms and learning (pp. 37-46). Oxford: Blackwell] repetition paradigm can be considered as a laboratory analogue of word learning. In Hebb learning experiments, the lists of items to be learned are presented as discrete sequences. In contrast, novel words are, by definition, always heard as a single coarticulated whole. Might this undermine the claim that Hebb learning can shed light on word learning? Here we report an experiment comparing learning sequences of isolated syllables with learning the same sequences spoken as a single coarticulated nonword. The pattern of learning was similar in the two cases, suggesting that the Hebb repetition paradigm can indeed provide valuable insights into the way novel word forms are learned.

Visual recency bias is explained by a mixture model of internal representations.

Human bias towards more recent events is a common and well-studied phenomenon. Recent studies in visual perception have shown that this recency bias persists even when past events contain no information about the future. Reasons for this suboptimal behavior are not well understood and the internal model that leads people to exhibit recency bias is unknown. Here we use a well-known orientation estimation task to frame the human recency bias in terms of incremental Bayesian inference. We show that the only Bayesian model capable of explaining the recency bias relies on a weighted mixture of past states. Furthermore, we suggest that this mixture model is a consequence of participants' failure to infer a model for data in visual short-term memory, and reflects the nature of the internal representations used in the task.

A shared representation of order between encoding and recognition in visual short-term memory.

Many complex tasks require people to bind individual events into a sequence that can be held in short term memory (STM). For this purpose information about the order of the individual events in the sequence needs to be maintained in an active and accessible form in STM over a period of few seconds. Here we investigated how the temporal order information is shared between the presentation and response phases of an STM task. We trained a classification algorithm on the fMRI activity patterns from the presentation phase of the STM task to predict the order of the items during the subsequent recognition phase. While voxels in a number of brain regions represented positional information during either presentation and recognition phases, only voxels in the lateral prefrontal cortex (PFC) and the anterior temporal lobe (ATL) represented position consistently across task phases. A shared positional code in the ATL might reflect verbal recoding of visual sequences to facilitate the maintenance of order information over several seconds.

Evidence accumulation in the integrated and primed Stroop tasks.

We report distributional analyses of response times (RT) in two variants of the color-word Stroop task using manual keypress responses. In the classic Stroop task, in which the color and word dimensions are integrated into a single stimulus, the Stroop congruence effect increased across the quantiles. In contrast, in the primed Stroop task, in which the distractor word is presented ahead of colored symbols, the Stroop congruence effect was manifested solely as a distributional shift, remaining constant across the quantiles. The distributional-shift pattern mirrors the semantic-priming effect that has been reported in semantic categorization tasks. The results are interpreted within the framework of evidence accumulation, and implications for the roles of task conflict and informational conflict are discussed.

Recall is not necessary for verbal sequence learning.

The question of whether overt recall of to-be-remembered material accelerates learning is important in a wide range of real-world learning settings. In the case of verbal sequence learning, previous research has proposed that recall either is necessary for verbal sequence learning (Cohen & Johansson Journal of Verbal Learning and Verbal Behavior, 6, 139-143, 1967; Cunningham, Healy, & Williams Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 575-597, 1984), or at least contributes significantly to it (Glass, Krejci, & Goldman Journal of Memory and Language, 28, 189-199, 1989; Oberauer & Meyer Memory, 17, 774-781, 2009). In contrast, here we show that the amount of previous spoken recall does not predict learning and is not necessary for it. We suggest that previous research may have underestimated participants' learning by using suboptimal performance measures, or by using manual or written recall. However, we show that the amount of spoken recall predicted how much interference from other to-be-remembered sequences would be observed. In fact, spoken recall mediated most of the error learning observed in the task. Our data support the view that the learning of overlapping auditory-verbal sequences is driven by learning the phonological representations and not the articulatory motor responses. However, spoken recall seems to reinforce already learned representations, whether they are correct or incorrect, thus contributing to a participant identifying a specific stimulus as either "learned" or "new" during the presentation phase.

Recalling visual serial order for verbal sequences.

We report three experiments in which participants performed written serial recall of visually presented verbal sequences with items varying in visual similarity. In Experiments 1 and 2 native speakers of Japanese recalled visually presented Japanese Kanji characters. In Experiment 3, native speakers of English recalled visually presented words. In all experiments, items varied in visual similarity and were controlled for phonological similarity. For Kanji and for English, performance on lists comprising visually similar items was overall poorer than for lists of visually distinct items across all serial positions. For mixed lists in which visually similar and visually distinct items alternated through the list, a clear "zig-zag" pattern appeared with better recall of the visually distinct items than for visually similar items. This is the first time that this zig-zag pattern has been shown for manipulations of visual similarity in serial-ordered recall. These data provide new evidence that retaining a sequence of visual codes relies on similar principles to those that govern the retention of a sequence of phonological codes. We further illustrate this by demonstrating that the data patterns can be readily simulated by at least one computational model of serial-ordered recall, the Primacy model (Page and Norris, Psychological Review, 105(4), 761-81, 1998). Together with previous evidence from neuropsychological studies and experimental studies with healthy adults, these results are interpreted as consistent with two domain-specific, limited-capacity, temporary memory systems for phonological material and for visual material, respectively, each of which uses similar processes that have evolved to be optimal for retention of serial order.

Bottoms up! How top-down pitfalls ensnare speech perception researchers, too.

Not only can the pitfalls that Firestone & Scholl (F&S) identify be generalised across multiple studies within the field of visual perception, but also they have general application outside the field wherever perceptual and cognitive processing are compared. We call attention to the widespread susceptibility of research on the perception of speech to versions of the same pitfalls.

The representation of order information in auditory-verbal short-term memory.

Here we investigate how order information is represented in auditory-verbal short-term memory (STM). We used fMRI and a serial recall task to dissociate neural activity patterns representing the phonological properties of the items stored in STM from the patterns representing their order. For this purpose, we analyzed fMRI activity patterns elicited by different item sets and different orderings of those items. These fMRI activity patterns were compared with the predictions made by positional and chaining models of serial order. The positional models encode associations between items and their positions in a sequence, whereas the chaining models encode associations between successive items and retain no position information. We show that a set of brain areas in the postero-dorsal stream of auditory processing store associations between items and order as predicted by a positional model. The chaining model of order representation generates a different pattern similarity prediction, which was shown to be inconsistent with the fMRI data. Our results thus favor a neural model of order representation that stores item codes, position codes, and the mapping between them. This study provides the first fMRI evidence for a specific model of order representation in the human brain.

Visual similarity effects on masked priming.

We investigated the role of the visual similarity of masked primes to targets in a lexical decision experiment. In the primes, some letters in the target (e.g., A in ABANDON) had either visually similar letters (e.g., H), dissimilar letters (D), visually similar digits (4), or dissimilar digits (6) substituted for them. The similarities of the digits and letters to the base letter were equated and verified in a two-alternative forced choice (2AFC) perceptual identification task. Using targets presented in lowercase (e.g., abandon) and primes presented in uppercase, visually similar digit primes (e.g., 484NDON) produced more priming than did visually dissimilar digit primes (676NDON), but little difference was found between the visually similar and dissimilar letter primes (HRHNDON vs. DWDNDON). These results were explained in terms of task-driven competition between the target letter and the visually similar letter.

Individual sequence representations in the medial temporal lobe.

Much of what we need to remember consists of sequences of stimuli, experiences, or events. Repeated presentation of a specific sequence establishes a more stable long-term memory, as shown by increased recall accuracy over successive trials of an STM task. Here we used fMRI to study the neural mechanisms that underlie sequence learning in the auditory-verbal domain. Specifically, we track the emergence of neural representations of sequences over the course of learning using multivariate pattern analysis. For this purpose, we use a serial recall task, in which participants have to recall overlapping sequences of letter names, with some of those sequences being repeated and hence learned over the course of the experiment. We show that voxels in the hippocampus come to encode the identity of specific repeated sequences although the letter names were common to all sequences in the experiment. These changes could have not been caused by changes in overall level of activity or to fMRI signal-to-noise ratios. Hence, the present results go beyond conventional univariate fMRI methods in showing a critical contribution of medial-temporal lobe memory systems to establishing long-term representations of verbal sequences.

Diacritic priming in novice readers of diacritics.

Recent masked priming studies investigating the recognition of letters with diacritics with native readers of the script have consistently yielded an asymmetric pattern of priming such that a base-letter prime without the diacritic speeds up the recognition of the letter with a diacritic almost as much as an identity prime, but not vice versa (e.g., á-Á ≦ a-Á, but a-A ≪á-A). Here we tested English readers unfamiliar with diacritics in a letter match task using Japanese kana and the vowel letters of the Latin alphabet, and found the asymmetry was reduced to a negligible level (á-Á ≦ a-Á, and a-A ≦ á-A) (Experiments 1 and 2). However, the diacritic novices showed the asymmetric pattern of priming like the diacritic experts when the task condition explicitly required the letters with and without diacritics to be distinguished (Experiments 3 and 4). These results are explained in terms of how expertise moderates an early letter identification process in interpreting a visual feature as noise, or a signal diagnostic of letter identity. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Neural mechanisms underlying the grouping effect in short-term memory.

Dividing auditory sequence into groups, or imposing rhythmic, tonal, or spatial structure during presentation, improves recall performance. Several competing computational models have been proposed to account for these effects, but little is known about the neural correlates of grouping and hence the representations that encode grouped sequences. The present study used functional magnetic resonance imaging (fMRI) to compare the auditory encoding of grouped and ungrouped lists of sub-span (six letters) and supra-span (nine letters) length in an immediate serial recall (ISR) task. Analysis of activation revealed an extensive premotor and prefrontal network, which was significantly less active when short-term memory (STM) span was exceeded during encoding. Only primary auditory cortex showed an increase in activation when memory span was exceeded. Comparison of activation for grouped and ungrouped lists showed that during the subspan phase bilateral planum temporale showed less activation for grouped stimuli, while during the supra-span phase supramarginal and inferior parietal areas were more active for grouped lists. The magnitude of both temporal and parietal activations predicted enhanced recall of grouped lists. Thus neural signatures of grouping seem to reflect more structured processing in parietal areas instead of reliance on perceptual-auditory processing in temporal regions.

Orthographic processing is universal; it's what you do with it that's different.

We agree with Frost that the variety of orthographies in the world's languages complicates the task of "cracking the orthographic code." Frost suggests that orthographic processing must therefore differ between orthographies. We suggest that the same basic orthographic processes are applied to all languages. Where languages differ is in what the reader must do with the results of orthographic processing.