EARSHOT: A Minimal Neural Network Model of Incremental Human Speech Recognition

Despite the lack of invariance problem (the many-to-many mapping between acoustics and percepts), human listeners experience phonetic constancy and typically perceive what a speaker intends. Most models of human speech recognition (HSR) have side-stepped this problem, working with abstract, idealized inputs and deferring the challenge of working with real speech. In contrast, carefully engineered deep learning networks allow robust, real-world automatic speech recognition (ASR). However, the complexities of deep learning architectures and training regimens make it difficult to use them to provide direct insights into mechanisms that may support HSR. In this brief article, we report preliminary results from a two-layer network that borrows one element from ASR, long short-term memory nodes, which provide dynamic memory for a range of temporal spans. This allows the model to learn to map real speech from multiple talkers to semantic targets with high accuracy, with human-like timecourse of lexical access and phonological competition. Internal representations emerge that resemble phonetically organized responses in human superior temporal gyrus, suggesting that the model develops a distributed phonological code despite no explicit training on phonetic or phonemic targets. The ability to work with real speech is a major advance for cognitive models of HSR.     

Does Partisan Bias Modulate Neural Processing of Political Information? An Analysis of the Neural Correlates of Corruption and Positive Messages

Daily worldwide newspapers print articles exposing government corruption. Yet these messages do not lead to a loss of votes for the corrupt parties. Sympathizers’ partisan bias, which respectively minimizes and maximizes corruption and positive messages of their own parties respectively, is widely considered the main cause of the loss of effectiveness of political communications. Despite the well-established existence of such bias when processing political information, little is known as to its psychological origin. Through the use of neuroscience (fMRI), this study explores the underlying brain mechanisms of negative (corruption) and positive political messages related to a conservative and a socialist Spanish political party, as well as the differences between their sympathizers. The findings reveal that negative (vs. positive) political messages exert the greatest neuroimaging impact on the electorate, as shown in aversive, risk, and disappointment-related brain regions. Interestingly, we show that there exists a main partisan bias against opposite parties (and not a positive bias toward one’s own party) that stems from a higher risk, ambiguity, and disbelief provoked by both positive and negative information about rival parties. Furthermore, this bias was more pronounced among conservative supporters. The current findings provide valuable insights for political parties to improve their communication campaigns.     

Truth Without Reference: The Use of Fictional Names

Topoi - Singular terms without referents are called empty or vacuous terms. But not all of them are equally empty. In particular, not all proper names that fail to name an existing object fail in...     

Does cognitive reflection mediate the math gender gap at university admission in Chile?

Social Psychology of Education - In Chile, a vast and persistent gender gap in math performance at university admission has negative consequences for women’s opportunities. International...     

Effect of social organisation on interspecific differences in overmarking behaviour of foals in African equids

Animal Cognition - Overmarking remains an unstudied topic in juvenile mammals. We have previously documented a very high rate of overmarking by foals in four captive African equid species: mountain...     

Satisfiability Testing for Boolean Formulas Using Δ-trees

The tree-based data structure of -tree for propositional formulas is introduced in an improved and optimised form. The -trees allow a compact representation for negation normal forms as well as for a number of reduction strategies in order to consider only those occurrences of literals which are relevant for the satisfiability of the input formula. These reduction strategies are divided into two subsets (meaning- and satisfiability-preserving transformations) and can be used to decrease the size of a negation normal form A at (at most) quadratic cost. The reduction strategies are aimed at decreasing the number of required branchings and, therefore, these strategies allow to limit the size of the search space for the SAT problem.