Fabien Mathy

Similarity-dissimilarity competition in disjunctive classification tasks

Typical disjunctive artificial classification tasks require participants to sort stimuli according to rules such as “x likes cars only when black and coupe OR white and SUV.” For categories like this, increasing the salience of the diagnostic dimensions has two simultaneous effects: increasing the distance between members of the same category and increasing the distance between members of opposite categories. Potentially, these two effects respectively hinder and facilitate classification learning, leading to competing predictions for learning. Increasing saliency may lead to members of the same category to be considered less similar, while the members of separate categories might be considered more dissimilar. This implies a similarity-dissimilarity competition between two basic classification processes. When focusing on sub-category similarity, one would expect more difficult classification when members of the same category become less similar (disregarding the increase of between-category dissimilarity); however, the between-category dissimilarity increase predicts a less difficult classification. Our categorization study suggests that participants rely more on using dissimilarities between opposite categories than finding similarities between sub-categories. We connect our results to rule- and exemplar-based classification models. The pattern of influences of within- and between-category similarities are challenging for simple single-process categorization systems based on rules or exemplars. Instead, our results suggest that either these processes should be integrated in a hybrid model, or that category learning operates by forming clusters within each category.

A rule-based presentation order facilitates category learning

We investigated the mechanisms by which concepts are learned from examples by manipulating the presentation order in which the examples were presented to subjects. We introduce the idea of a rule-based presentation order, which is a sequence that respects the internal organization of the examples within a category. We find that such an order substantially facilitates learning, as compared with previously known beneficial orders, such as a similarity-based order. We discuss this result in light of the central distinction between rule-based and similaritybased learning models.

A snapshot is all it takes to encode object locations into spatial memory

This study examines the encoding of multiple object locations into spatial memory by comparing localization accuracy for stimuli presented at different exposure durations. Participants in the longest duration condition viewed masked displays containing 1-10 discs for 1-10 s (durations typically used in simple span tasks), and then reported the locations of these discs on a blank screen. Compared to conditions that presented the same stimuli briefly for 50 or 200 ms (exposures more typical of simultaneous spatial arrays), localization accuracy did not improve significantly under longer viewing durations. Additionally, a clustering analysis found that responses were spread among different clusters of discs and not focused on individual clusters, regardless of viewing duration. A second experiment tested this performance for displays containing two distinct clusters of discs to determine if clearly grouped subsets of objects would improve performance, but there was no substantial improvement for these two-cluster displays when compared to displays with one cluster. Overall, the results indicate that spatial information for a set of objects is extracted globally and quickly, with little benefit from extended encoding durations that should have favored some deliberative form of grouping. Such results cast doubt on the validity of Corsi blocks or equivalent common neuropsychological tests purportedly designed to evaluate specifically spatial short-term memory spans.

Chunk formation in immediate memory and how it relates to data compression

This paper attempts to evaluate the capacity of immediate memory to cope with new situations in relation to the compressibility of information likely to allow the formation of chunks. We constructed a task in which untrained participants had to immediately recall sequences of stimuli with possible associations between them. Compressibility of information was used to measure the chunkability of each sequence on a single trial. Compressibility refers to the recoding of information in a more compact representation. Although compressibility has almost exclusively been used to study long-term memory, our theory suggests that a compression process relying on redundancies within the structure of the list materials can occur very rapidly in immediate memory. The results indicated a span of about three items when the list had no structure, but increased linearly as structure was added. The amount of information retained in immediate memory was maximal for the most compressible sequences, particularly when information was ordered in a way that facilitated the compression process. We discuss the role of immediate memory in the rapid formation of chunks made up of new associations that did not already exist in long-term memory, and we conclude that immediate memory is the starting place for the reorganization of information.

The long-term effect of relational information in classification learning

This study examines the long-term effect of mutual information in the learning of Shepardian classifications. Mutual information is a measure of the complexity of the relationship between features because it quantifies how the features relate to each other. For instance, in various categorisation models, Type VI concepts—originally studied by Shepard, Hovland, and Jenkins (1961)—are unanimously judged to be the most complex kind of 3-D Boolean concepts. This has been largely confirmed by empirical data. Yet, it is apparently inconsistent with the fact that this concept entails the greatest amount of mutual information of all the 3-D Boolean concepts. The present study was aimed at verifying whether individuals can use relational information, in the long run, to devise easier strategies for category learning. Subject performance was measured repeatedly for 1 hour on either successive Type VI concepts (using different features between problems) or successive Type IV concepts. The results showed that shortly after the second problem, Type VI concepts became easier to learn than Type IV ones. The gap between the mean per-problem error rates of the two concepts continued to increase as the number of problems increased. Two other experiments tended to confirm this trend. The discussion brings up the idea of combining different metrics in categorisation models in order to include every possible way for subjects to simplify the categorisation process.

An extended study of the nonindependence of stimulus properties in human classification learning

Categorization researchers have tried to verify their models through laboratory experiments with simplified stimulus sets, a requirement that can rarely be met in real-world situations in which properties are often connected. Still, the targeted simplification of the material might be illusory. We replicate and extend Love and Markmans (2003) study of the nonindependence of canonical stimulus properties such as size, colour, and shape in human classification learning, in which the authors concluded that shape takes precedence over other dimensions. To support their hypothesis, Love and Markman showed that certain classifications are more difficult for participants when shape is combined to one of its putative subordinate features, size or colour, than when shape is irrelevant to the task. A data set of 290 → 50 adult participants completing one or more classification tasks was collected. The results confirm that certain combinations of shape, size, and colour can hinder or facilitate classification learning, but not necessarily in the form expected by the nonindependence postulated by Love and Markman, especially in Experiment 2 where a totally reverse pattern of difficulty is observed (shape does not take precedence over other dimensions). Also, we show that simple similarity effects in clustering retain considerable intuitive appeal and can offer an alternative account to the nonindependence of stimulus properties, especially because slight variations in the dimensions chosen make the observations of Love and Markman unstable.

Mathematical transcription of the ‘time-based resource sharing’ theory of working memory

The time-based resource sharing (TBRS) model is a prominent model of working memory that is both predictive and simple. TBRS is a mainstream decay-based model and the most susceptible to competition with interference-based models. A connectionist implementation of TBRS, TBRS*, has recently been developed. However, TBRS* is an enriched version of TBRS, making it difficult to test general characteristics resulting from TBRS assumptions. Here, we describe a novel model, TBRS2, built to be more transparent and simple than TBRS*. TBRS2 is minimalist and allows only a few parameters. It is a straightforward mathematical transcription of TBRS that focuses exclusively on the activation level of memory items as a function of time. Its simplicity makes it possible to derive several theorems from the original TBRS and allows several variants of the refresh process to be tested without relying on particular architectures.

Inner speech sustains predictable task switching: direct evidence in adults

ABSTRACT It has been proposed that inner speech supports task selection in task-switching studies, especially when the need for endogenous control is increased. This has been established through the suppression of inner speech in cognitive-flexibility tasks that leads to poorer performance. The aim of this study is to quantify the role of inner speech in a flexibility task by using surface laryngeal electromyography, which, contrary to previous studies, enables participants to freely verbalise the tasks. We manipulated endogenous and exogenous flexibility in a mathematical switching task paradigm. Experiment 1 shows that inner speech acts as a support for switching and is recruited more often when the tasks are of an endogenous type. The main result of Experiment 2 that language is recruited more for the mixing cost than for the switch cost (regardless of the endogenous factor) extends past findings obtained through articulatory suppression.

Developmental Abilities to Form Chunks in Immediate Memory and Its Non-Relationship to Span Development

Both adults and children –by the time they are 2–3 years old– have a general ability to recode information to increase memory efficiency. This paper aims to evaluate the ability of untrained children aged 6–10 years old to deploy such a recoding process in immediate memory. A large sample of 374 children were given a task of immediate serial report based on SIMON®, a classic memory game made of four colored buttons (red, green, yellow, blue) requiring players to reproduce a sequence of colors within which repetitions eventually occur. It was hypothesized that a primitive ability across all ages (since theoretically already available in toddlers) to detect redundancies allows the span to increase whenever information can be recoded on the fly. The chunkable condition prompted the formation of chunks based on the perceived structure of color repetition within to-be-recalled sequences of colors. Our result shows a similar linear improvement of memory span with age for both chunkable and non-chunkable conditions. The amount of information retained in immediate memory systematically increased for the groupable sequences across all age groups, independently of the average age-group span that was measured on sequences that contained fewer repetitions. This result shows that chunking gives young children an equal benefit as older children. We discuss the role of recoding in the expansion of capacity in immediate memory and the potential role of data compression in the formation of chunks in long-term memory.

The Influence of Presentation Order on Category Transfer.

This study of supervised categorization shows how different kinds of category representations are influenced by the order in which training examples are presented. We used the well-studied 5-4 category structure of Medin and Schaffer (1978) , which allows transfer of category learning to new stimuli to be discriminated as a function of rule-based or similarity-based category knowledge. In the rule-based training condition (thought to facilitate the learning of abstract logical rules and hypothesized to produce rule-based classification), items were grouped by subcategories and randomized within each subcategory. In the similarity-based training condition (thought to facilitate associative learning and hypothesized to produce exemplar classification), transitions between items within the same category were determined by their featural similarity and subcategories were ignored. We found that transfer patterns depended on whether the presentation order was similarity-based, or rule-based, with the participants particularly capitalizing on the rule-based order.