Search

How we judge the similarity between objects in the world is connected ultimately to how we represent those objects. It has been argued extensively that object representations in humans are ‘structured’ in nature, meaning that both individual features and the relations between them can influence similarity. In contrast, popular models within comparative psychology assume that nonhuman species appreciate only surface-level, featural similarities. By applying psychological models of structural and featural similarity (from conjunctive feature models to Tversky’s contrast model) to visual similarity judgements from adult humans, chimpanzees, and gorillas, we demonstrate a cross-species sensitivity to complex structural information, particularly for stimuli that combine colour and shape. These results shed new light on the representational complexity of nonhuman apes, and the fundamental limits of featural coding in explaining object representation and similarity, which emerge strikingly across both human and nonhuman species.

The idea of resolving dispute through the exchange of arguments and reasons has been central to society for millennia. We exchange arguments as a way of getting at the truth in contexts as diverse as science, the court room, and our everyday lives. In democracies, political decisions should be negotiated through argument, not deception, or even worse, brute force. If argument is to lead to the truth or to good decisions, then some arguments must be better than others and 'argument strength' must have some meaningful connection with truth. Can argument strength be measured in a way that tracks an objective relationship with truth and not just mere persuasiveness? This article describes recent developments in providing such measures.

Counterfactual (CF) explanations have been employed as one of the modes of explainability in explainable artificial intelligence (AI)-both to increase the transparency of AI systems and to provide recourse. Cognitive science and psychology have pointed out that people regularly use CFs to express causal relationships. Most AI systems, however, are only able to capture associations or correlations in data, so interpreting them as casual would not be justified. In this perspective, we present two experiments (total n = 364) exploring the effects of CF explanations of AI systems' predictions on lay people's causal beliefs about the real world. In Experiment 1, we found that providing CF explanations of an AI system's predictions does indeed (unjustifiably) affect people's causal beliefs regarding factors/features the AI uses and that people are more likely to view them as causal factors in the real world. Inspired by the literature on misinformation and health warning messaging, Experiment 2 tested whether we can correct for the unjustified change in causal beliefs. We found that pointing out that AI systems capture correlations and not necessarily causal relationships can attenuate the effects of CF explanations on people's causal beliefs.

Base rate neglect refers to people's apparent tendency to underweight or even ignore base rate information when estimating posterior probabilities for events, such as the probability that a person with a positive cancer-test outcome actually does have cancer. While often replicated, almost all evidence for the phenomenon comes from studies that used problems with extremely low base rates, high hit rates, and low false alarm rates. It is currently unclear whether the effect generalizes to reasoning problems outside this "corner" of the entire problem space. Another limitation of previous studies is that they have focused on describing empirical patterns of the effect at the group level and not so much on the underlying strategies and individual differences. Here, we address these two limitations by testing participants on a broader problem space and modeling their responses at a single-participant level. We find that the empirical patterns that have served as evidence for base-rate neglect generalize to a larger problem space, albeit with large individual differences in the extent with which participants "neglect" base rates. In particular, we find a bi-modal distribution consisting of one group of participants who almost entirely ignore the base rate and another group who almost entirely account for it. This heterogeneity is reflected in the cognitive modeling results: participants in the former group were best captured by a linear-additive model, while participants in the latter group were best captured by a Bayesian model. We find little evidence for heuristic models. Altogether, these results suggest that the effect known as "base-rate neglect" generalizes to a large set of reasoning problems, but varies largely across participants and may need a reinterpretation in terms of the underlying cognitive mechanisms.

There has been much interest in group judgment and the so-called ‘wisdom of crowds’. In many real world contexts, members of groups not only share a dependence on external sources of information, but they also communicate with one another, thus introducing correlations among their responses that can diminish collective accuracy. This has long been known, but it has—to date—not been examined to what extent different kinds of communication networks may give rise to systematically different effects on accuracy. We argue that equations that relate group accuracy, individual accuracy, and group diversity (see Hogarth, 1978; Page, 2007) are useful theoretical tools for understanding group performance in the context of research on group structure. In particular, these equations may serve to identify the kind of group structures that improve individual accuracy without thereby excessively diminishing diversity so that the net positive effect is an improvement even on the level of collective accuracy. Two experiments are reported where two structures (the complete network and a small world network) are investigated from this perspective. It is demonstrated that the more constrained network (the small world network) outperforms the network with a free flow of information.

Slippery slope arguments (SSAs) have a bad philosophical reputation. They seem, however, to be widely used and frequently accepted in many legal, political, and ethical contexts. Hahn and Oaksford (2007) argued that distinguishing strong and weak SSAs may have a rational basis in Bayesian decision theory. In this paper three experiments investigated the mechanism of the slippery slope showing that they may have an objective basis in category boundary re-appraisal. When the beginning and the end of a slippery slope are more similar, the probability that they are perceived to belong in the same category is higher and the SSA is stronger. Experiment 1 established a robust effect of probability on SSA evaluation. Experiments 2 and 2A showed that when similar items are classified in the same category this leads to stronger SSAs. In Experiment 3, in a correlational analysis, it was shown that participants’ confidence in their categorisation judgements predicted the perceived strength of an SSA and that this relationship was moderated by similarity between the ends of the slippery slope. We conclude that an important aspect of many SSAs may have an objective basis in well-established and rational cognitive theories.

This paper contrasts two structural accounts of psychological similarity: structural alignment (SA) and Representational Distortion (RD). SA proposes that similarity is determined by how readily the structures of two objects can be brought into alignment; RD measures similarity by the complexity of the transformation that "distorts" one representation into the other. We assess RD by defining a simple coding scheme of psychological transformations for the experimental materials. In two experiments, this "concrete" version of RD provides compelling fits of the data and compares favourably with SA. Finally, stepping back from particular models, we argue that perceptual theory suggests that transformations and alignment processes should generally be viewed as complementary, in contrast to the current distinction in the literature.

The human anthrax vaccines currently licensed contain the protective antigen (PA) of Bacillus anthracis as main antigen together with traces of some other bacillus components, e.g. lethal factor (LF). The present study aimed at monitoring the course of specific antibody titres against PA and LF by enzyme linked immunosorbent assays (ELISA), as well as the levels of toxin-neutralising antibodies, in 11 volunteers vaccinated with the human anthrax vaccine UK. After an initial seroconversion in all vaccinees, a significant reduction of both antibody titres against PA and LF, and of neutralising antibodies, was detected just prior to a vaccine boost 6 months after completion of the basic immunisation. Following the booster injection, titres increased again to levels comparable to those after the fourth immunisation. ELISA titres against PA correlated significantly with neutralising antibodies (r=0.816, p

Although similarity plays an important role in accounts of language processing, there are surprisingly few direct empirical studies of the phonological similarity between words, and it is therefore not clear whether similarity comparisons between words involve processes similar to those involved in other cognitive domains. In five experiments, participants chose which of two monosyllabic pseudo-words sounded more similar to a target pseudo-word. Our results are generally consistent with the structural alignment theory of comparisons between complex mental representations, suggesting that phonological word similarity parallels similarity involving other kinds of information including visual objects and scenes, events, and word meanings. We use our results to test new metrics of word similarity, and identify predictions for future similarity research both in the domain of word sounds and in other cognitive domains.

Similarity between component speech sounds influences language processing in numerous ways. Explanation and detailed prediction of linguistic performance consequently requires an understanding of these basic similarities. The research reported in this paper contrasts two broad classes of approach to the issue of phoneme similarity—theoretically driven measures based on phonological features, and empirically derived measures based on confusability. Two experiments are used to test a variety of measures from both classes for their ability to predict judgments of similarity between English consonants. The simplest featural measure is the best on these tests. This recommends it as the basis for further explorations, and improvements to the basic feature model are identified. The paper concludes by examining the extent to which confusability depends on phoneme similarity.

Wordlikeness, the extent to which a sound sequence is typical of words in a language, affects language acquisition, language processing, and verbal short-term memory. Wordlikeness has generally been equated with phonotactic knowledge of the possible or probable sequences of sounds within a language. Alternatively, wordlikeness might be derived directly from the mental lexicon, depending only on similarity to known words. This paper tests these two cognitively different possibilities by comparing measures of phonotactic probability and lexical influence, including a new model of lexical neighborhoods, in their ability to explain empirical wordlikeness judgments. Our data show independent contributions of both phonotactic probability and the lexicon, with relatively greater influence from the lexicon. The influence of a lexical neighbor is found to be an inverted-Ushaped function of its token frequency. However, our results also indicate that current measures are limited in their ability to account for sequence typicality.

The distinction between rule-based and similarity-based processes in cognition is of fundamental importance for cognitive science, and has been the focus of a large body of empirical research. However, intuitive uses of the distinction are subject to theoretical difficulties and their relation to empirical evidence is not clear. We propose a 'core' distinction between rule- and similarity-based processes, in terms of the way representations of stored information are 'matched' with the representation of a novel item. This explication captures the intuitively clear-cut cases of processes of each type, and resolves apparent problems with the rule/similarity distinction. Moreover, it provides a clear target for assessing the psychological and AI literatures. We show that many lines of psychological evidence are less conclusive than sometimes assumed, but suggest that converging lines of evidence may be persuasive. We then argue that the AI literature suggests that approaches which combine rules and similarity are an important new focus for empirical work.

The predominant antigen in human anthrax vaccines is the protective antigen (PA) of Bacillus anthracis. To address the question whether immune responses against B. anthracis spores can improve the protectivity of a PA-based DNA vaccine against anthrax, we designed a eucaryotic expression plasmid encoding the exosporium antigen BclA, which is a collagen-like surface protein. This plasmid, pSecTag BclA, carries a secretion signal for the recombinant antigen. Using NMRI mice we compared the effects of immunisation with a combination of PA- and BclA-encoding plasmid DNA, to immunisations with either PA- or BclA-encoding pDNA. After three immunisations the mice were infected with 25 × LD50 of B. anthracis Ames spores. The plasmid pSecTag BclA, induced high BclA-specific antibody responses. Vaccination with a combination of PA- and BclA-encoding pDNA led to significantly better survival than immunisation with only PA- or only BclA-encoding plasmids.

Voltage-gated K+ (Kv) channels play a crucial role in the generation of cellular action potentials. Kv channels open and close in response to changes in transmembrane voltage. A further regulatory mechanism is channel inactivation, which occurs at the K+ selectivity filter and is of particular importance in the cardiac K+ channel hERG, that plays a major role in controlling the human heart rate. In a series of molecular dynamics simulations, we identified an extensive hydrogen bond network around residues S620 and N629 that controls the conformation of the selectivity filter and should therefore have a great impact on inactivation. We show that the distance of N629 to S620 acts as a nearly linear switch to drive the K+ selectivity filter from its conductive to a collapsed state similar to the low K+ conformation of KcsA. Electrophysiological measurements on mutants with altered H-bonding networks display either reduced or even fully abolished inactivation and thus support our hypothesis. Additionally, our simulations revealed a sidepocket extending from the central cavity, only present in the conductive state. These pockets are sufficently large to accommodate molecules of the newly found class of hERG-activators and reveal a first hint at their putative molecular mechanism.