Esther Mondragón

City, University of London · Department of Computer Science

Using rules extracted from experience to solve problems in novel situations involves cognitions such as analogical reasoning and language learning and is considered a keystone of humans' unique abilities. Nonprimates, it has been argued, lack such rule transfer. We report that Rattus norvegicus can learn simple rules and apply them to new situation...

This paper presents a novel representational framework for the Temporal Difference (TD) model of learning, which allows the computation of configural stimuli – cumulative compounds of stimuli that generate perceptual emergents known as configural cues. This Simultaneous and Serial Configural-cue Compound Stimuli Temporal Difference model (SSCC TD)...

Non-reinforced preexposure to two stimuli often enhances discrimination between them. Analyses of this perceptual learning phenomenon have mainly focused on the role played by the distinctive stimulus features; this study examined the contribution of the non-distinctive common elements. A standard appetitive Pavlovian procedure was used. Rats recei...

In this paper we present the "R&W Simulator" (version 3.0), a Java simulator of Rescorla and Wagner's prediction error model of learning. It is able to run whole experimental designs, and compute and display the associative values of elemental and compound stimuli simultaneously, as well as use extra configural cues in generating compound values; i...

This article explores the contribution of the double error dynamic asymptote computational associative learning model to understanding the role of mediated learning mechanisms in the generation of spurious associations, as those postulated to characterize schizophrenia. Three sets of simulations for mediated conditioning, mediated extinction, and a...

Inspired by cognitive theories of creativity, this paper introduces a computational model (AIGenC) that lays down the necessary components to enable artificial agents to learn, use and generate transferable representations. Unlike machine representations, which rely exclusively on raw sensory data, biological representations incorporate relational...

In this paper, we propose a framework to extract the algebra of the transformations of worlds from the perspective of an agent. As a starting point, we use our framework to reproduce the symmetry-based representations from the symmetry-based disentangled representation learning (SBDRL) formalism proposed by [1]; only the algebra of transformations...

This paper introduces a computational model of creative problem-solving in deep reinforcement learning agents, inspired by cognitive theories of creativity. The AIGenC model aims at enabling artificial agents to learn, use and generate transferable representations. AIGenC is embedded in a deep learning architecture that includes three main componen...

This paper introduces a computational model of creative problem-solving in deep reinforcement learning agents, inspired by cognitive theories of creativity. The AIGenC model aims at enabling artificial agents to learn, use and generate transferable representations. AIGenC is embedded in a deep learning architecture that includes three main componen...

Multi-agent reinforcement learning in mixed-motive settings allows for the study of complex dynamics of agent interactions. Embodied agents in partially observable environments with the ability to communicate can share information, agree on strategies, or even lie to each other. In order to study this, we propose a simple environment where we can i...

In this article a formal model of associative learning is presented that incorporates representational and computational mechanisms that, as a coherent corpus, empower it to make accurate predictions of a wide variety of phenomena that, so far, have eluded a unified account in learning theory. In particular, the Double Error Dynamic Asymptote (DDA)...

In this paper a formal model of associative learning is presented which incorporates representational and computational mechanisms that, as a coherent corpus, empower it to make accurate predictions of a wide variety of phenomena that so far have eluded a unified account in learning theory. In particular, the Double Error Dynamic Asymptote (DDA) mo...

Computational models of classical conditioning have made significant contributions to the theoretic understanding of associative learning, yet they still struggle when the temporal aspects of conditioning are taken into account. Interval timing models have contributed a rich variety of time representations and provided accurate predictions for the...

In this paper a formal model of associative learning is presented which incorporates representational and computational mechanisms that, as a coherent corpus, empower it to make accurate predictions of a wide variety of phenomena that so far have eluded a unified account in learning theory. In particular, the Double Error model introduces: 1) a ful...

Conditioning, how animals learn to associate two or more events, is one of the most influential paradigms in learning theory. It is nevertheless unclear how current models of associative learning can accommodate complex phenomena without ad hoc representational assumptions. We propose to embrace deep neural networks to negotiate this problem.

Book of Abstracts of the First Workshop on Computational Modelling of Classical Conditioning. CoMoCC, London 2016

This report introduces a formal comparison between Hall and Rodríguez’s elaboration of the Pearce and Hall model and the original. It shows mathematically and with a set of simulations that with a provision to set an initial excitatory value in Pearce and Hall’s algorithm, both models are equivalent in structure and that they result in identical pr...

In a typical conditioning task, a conditioned stimulus (CS) is reliably followed by an outcome of motivational value. As a result, a conditioned response (CR) develops during the CS, indicating anticipation of the unconditioned stimulus (US). This chapter considers the temporal characteristics of this process, and examines the extent to which they...

Whether animals behave optimally is an open question of great importance, both theoretically and in practice. Attempts to answer this question focus on two aspects of the optimization problem, the quantity to be optimized and the optimization process itself. In this paper, we assume the abstract concept of cost as the quantity to be minimized and p...

Abstract Two experiments investigated the effect of the temporal distribution form of a stimulus on its ability to produce an overshadowing effect. The overshadowing stimuli were either of the same duration on every trial, or of a variable duration drawn from an exponential distribution with the same mean duration as that of the fixed stimulus. Bot...

The last 50 years have seen the progressive refinement of our understanding of the mechanisms of classical conditioning and this has resulted in the development of several influential theories that are able to explain with considerable precision a wide variety of experimental findings, and to make non-intuitive predictions that have been confirmed....

Classical conditioning is a fundamental paradigm in the study of learning and thus in understanding cognitive processes and behaviour, for which we need comprehensive and accurate models. This paper aims at evaluating and comparing a collection of influential computational models of classical conditioning by analysing the models themselves and agai...

An implicit assumption in the study of operant conditioning and reinforcement learning is that behavior is stochastic, in that it depends on the probability that an outcome follows a response and on how the presence or absence of the output affects the frequency of the response. In this paper we argue that classical probability is not the right too...

In four experiments rats were conditioned to an auditory conditioned stimulus (conditioned stimulus; CS) that was paired with food, and learning about the CS was compared across two conditions in which the mean duration of the CS was equated. In one, the CS was of a single, fixed duration on every trial, and in the other the CS duration was drawn f...

In this position paper we propose to enhance learning algorithms, reinforcement learning in particular, for agents and for multi-agent systems, with the introduction of concepts and mechanisms borrowed from associative learning theory. It is argued that existing algorithms are limited in that they adopt a very restricted view of what "learning" is,...

It is well known that, in one form or another, the variational Principle of Least Action (PLA) governs Nature. Although traditionally referred to explain physical phenomena, PLA has also been used to account for biological phenomena and even natural selection. However, its value in studying psychological processes has not been fully explored. In th...

The authors propose in this chapter to use abstract algebra to unify different models of theories of associative learning -- as complementary to current psychological, mathematical and computational models of associative learning phenomena and data. The idea is to compare recent research in associative learning to identify the symmetries of behavio...

Standard associative learning theories typically fail to conceptualise the temporal properties of a stimulus, and hence cannot easily make predictions about the effects such properties might have on the magnitude of conditioning phenomena. Despite this, in intuitive terms we might expect that the temporal properties of a stimulus that is paired wit...

The acquisition of a negative evaluation of a fictitious minority social group in spite of the absence of any objective correlation between group membership and negative behaviours was described by Hamilton and Gifford (1976) as an instance of an illusory correlation. We studied the acquisition and attenuation through time of this correlation learn...

In recent years there has been increased interest in developing computational and mathematical models of learning and adaptation. Computational Neuroscience for Advancing Artificial Intelligence: Models, Methods and Applications captures the latest research in this area, providing a learning theorists with a mathematically sound framework within wh...

Standard associative learning theories typically fail to conceptualise the temporal properties of a stimulus, and hence cannot easily make predictions about the effects such properties might have on the magnitude of conditioning phenomena. Despite this, in intuitive terms we might expect that the temporal properties of a stimulus that is paired wit...

The authors propose in this chapter to use abstract algebra to unify different models of theories of associative learning -- as complementary to current psychological, mathematical and computational models of associative learning phenomena and data. The idea is to compare recent research in associative learning to identify the symmetries of behavio...

The critical review of the rats learning pattern XYX and its transfer to novel stimuli is presented. Corballis claims that human infants and cotton top tamarins, if confronted with the same kind of task, may have used a subset of stimuli to solve the rule discrimination. Rats learning that XYX was the reinforced sequence may have matched the identi...

Theories of causal cognition describe how animals code cognitive primitives such as causal strength, directionality of relations, and other variables that allow inferences on the effect of interventions on causal links. We argue that these primitives and importantly causal generalization can be studied within an animal learning framework. Causal ma...

The serial order in which events occur can be a signal for different outcomes and therefore might be a determinant of how an animal should respond. In this report, we propose a novel design for studying serial order learning in Pavlovian conditioning. In both Experiments 1a and 1b, hungry rats were trained with successively presented pairs of audit...

Rats received training in which two auditory target stimuli, X and Y, were signaled by two visual stimuli, A and B, and followed by food (i.e., A-->X+, B-->Y+). The test consisted of presentations of X and Y preceded either by the same signal as during training (same trials: A-->X, B-->Y) or by the alternative signal (different trials: A-->Y, B-->X...

In this paper we contend that adaptation and learning are essential in designing and building autonomous software systems for real-life applications. In particular, we will argue that in dynamic, complex domains autonomy and adaptability go hand by hand, that is, that agents cannot make their own decisions if they are not provided with the ability...

Rats received exposure to two compound flavours, AX and BX, where A and B were sucrose and saline and X was acid. For group intermixed (1), exposure consisted of alternating trials with AX and BX; group blocked (B) received a block of AX trials and a separate block of BX trials. Experiment 1 showed that generalization to BX after conditioning with...