Irina Baetu

Human judgments of positive and negative causal chains

Three experiments investigated the way participants construct causal chains from experience with the individual links that make up those chains. Participants were presented with contingency information about the relationship between events A and B, as well as events B and C, using trial-by-trial presentations. The A-B and B-C contingencies could be positive, negative, or zero. Although participants had never experienced A and C together, A-C ratings were a multiplicative function of the A-B and B-C contingencies. These findings can be generated by an auto-associator using the delta rule. This explanation is also useful for understanding sensory preconditioning and second-order conditioning.

Extinction and blocking of conditioned inhibition in human causal learning.

Two experiments investigated extinction and blocking of a conditioned inhibitor in a human contingency learning task. Lotz and Lachnit (2009) and Melchers, Wolff, and Lachnit (2006) reported extinction of inhibition only when participants experienced outcome levels lower than those used in training. In Experiment 1, which used more neutral instructions than the previously mentioned studies, we found that extinction of inhibition occurred, whether or not participants experienced lower outcome levels. In Experiment 2, we applied this outcome manipulation to blocking of a conditioned inhibitor. We found blocking of inhibition both when participants had experience with lower outcomes and when they did not. The results of our two experiments are consistent with Rescorla and Wagner’s (1972) associative model, and inconsistent with an inferential account of causal learning (De Houwer, Beckers, & Vandorpe, 2005).

A comparative approach to cue competition with one and two strong predictors.

The relative validity effect (Wagner, Logan, Haberlandt, & Price, 1968) demonstrated that a strong cue or cause reduces responding to, or judgments of, a weaker cue or cause. We report two experiments with human subjects using relative validity preparations in which we investigate one- and two-cue competition effects. Previously, we investigated the effect using instrumental and Pavlovian conditioning preparations with rats. In the first experiment, we used a procedure analogous to the animal preparations. In the second experiment, we used a different probabilistic procedure. The results with humans and rats are very similar. In each species we find similar interference with processing the moderate predictor with one or with two strong competitors.

Commonly-occurring polymorphisms in the COMT, DRD1 and DRD2 genes influence different aspects of motor sequence learning in humans

Performing sequences of movements is a ubiquitous skill that involves dopamine transmission. However, it is unclear which components of the dopamine system contribute to which aspects of motor sequence learning. Here we used a genetic approach to investigate the relationship between different components of the dopamine system and specific aspects of sequence learning in humans. In particular, we investigated variations in genes that code for the catechol-O-methyltransferase (COMT) enzyme, the dopamine transporter (DAT) and dopamine D1 and D2 receptors (DRD1 and DRD2). COMT and the DAT regulate dopamine availability in the prefrontal cortex and the striatum, respectively, two key regions recruited during learning, whereas dopamine D1 and D2 receptors are thought to be involved in long-term potentiation and depression, respectively. We show that polymorphisms in the COMT, DRD1 and DRD2 genes differentially affect behavioral performance on a sequence learning task in 161 Caucasian participants. The DRD1 polymorphism predicted the ability to learn new sequences, the DRD2 polymorphism predicted the ability to perform a previously learnt sequence after performing interfering random movements, whereas the COMT polymorphism predicted the ability to switch flexibly between two sequences. We used computer simulations to explore potential mechanisms underlying these effects, which revealed that the DRD1 and DRD2 effects are possibly related to neuroplasticity. Our prediction-error algorithm estimated faster rates of connection strengthening in genotype groups with presumably higher D1 receptor densities, and faster rates of connection weakening in genotype groups with presumably higher D2 receptor densities. Consistent with current dopamine theories, these simulations suggest that D1-mediated neuroplasticity contributes to learning to select appropriate actions, whereas D2-mediated neuroplasticity is involved in learning to inhibit incorrect action plans. However, the learning algorithm did not account for the COMT effect, suggesting that prefrontal dopamine availability might affect sequence switching via other, non-learning, mechanisms. These findings provide insight into the function of the dopamine system, which is relevant to the development of treatments for disorders such as Parkinsons disease. Our results suggest that treatments targeting dopamine D1 receptors may improve learning of novel sequences, whereas those targeting dopamine D2 receptors may improve the ability to initiate previously learned sequences of movements.

Accuracy-based measures provide a better measure of sequence learning than reaction time-based measures

The Serial Reaction Time Task (SRTT) was designed to measure motor sequence learning and is widely used in many fields in cognitive science and neuroscience. However, the common performance measures derived from SRTT—reaction time (RT) difference scores—may not provide valid measures of sequence learning. This is because RT-difference scores may be subject to floor effects and otherwise not sufficiently reflective of learning. A ratio RT measure might minimize floor effects. Furthermore, measures derived from predictive accuracy may provide a better assessment of sequence learning. Accordingly, we developed a Predictive Sequence Learning Task (PSLT) in which performance can be assessed via both RT and predictive accuracy. We compared performance of N = 99 adults on SRTT and PSLT in a within-subjects design and also measured fluid abilities. The RT-difference scores on both tasks were generally not related to fluid abilities, replicating previous findings. In contrast, a ratio RT measure on SRTT and PSLT and accuracy measures on PSLT were related to fluid abilities. The accuracy measures also indicated an age-related decline in performance on PSLT. The current patterns of results were thus inconsistent across different measures on the same tasks, and we demonstrate that this discrepancy is potentially due to floor effects on the RT difference scores. This may limit the potential of SRTT to measure sequence learning and we argue that PSLT accuracy measures could provide a more accurate reflection of learning ability.

Competition between multiple causes of a single outcome in causal reasoning

A strong positive predictor of an outcome modulates the causal judgments of a moderate predictor. To study the empirical basis of this modulation, we compared treatments with one and with two strong competing (i.e., modulating) causes. This allowed us to vary the frequency of outcome occurrences or effects paired with the predictors. We investigated causal competition between positive predictors (those signaling the occurrence of the outcome), between negative predictors (those signaling the absence of the outcome) and between predictors of opposite polarity (positive and negative). The results are consistent with a contrast rather than a reduced associative strength or conditional contingency account, because a strong predictor of opposite polarity enhances rather than reduces causal estimates of moderate predictors. In addition, we found competition effects when the strong predictor predicted fewer outcome occurrences than the moderate predictor, thus implying that cue competition is, at least sometimes, a consequence of contingency rather than total cue-outcome pairings.

Mental models of causation: A comparative view

Blocking effects and associations Contiguity or the pairing of events has long been recognized by learning theorists as insufficient to explain basic associative processes. In 1968, Leon Kamin described the blocking phenomenon as a demonstration of this insufficiency (Kamin, 1968). Using a two-phase design, as shown in Table 4.1, Kamin conditioned a group of animals to associate a single cue with an outcome (i.e., A Æ O). In Phase 2, a second cue was paired alongside the first (i.e., AB Æ O). AContents: Preface. A.J. Wills, Association and Cognition. A.G. Baker, R. Murphy, R. Mehta, I. Baetu, Mental Models of Causation: A Comparative View. J. De Houwer, S. Vandorpe, T. Beckers, On the Role of Controlled Cognitive Processes in Human Associative Learning. J.M. Tangen, L.G. Allan, H. Sadeghi, Assessing (In)sensitivity to Casual Asymmetry: A Matter of Degree. A.J. Wills, Connectionist Models of Human Associative Learning. J. Zwickel, A.J. Wills, Integrating Associative Models of Supervised and Unsupervised Categorization. M.E. Le Pelley, I.P.L. McLaren, The Role of Associative History in Human Casual Learning. M. Suret, I.P.L. McLaren, Elemental Representation and Associability: An Integrated Model. A.J. Wills, Applications and Extensions. R.A. Murphy, F. Vallee-Tourangeau, R. Msetfi, A.G. Baker, Signal-Outcome Contingency, Contiguity, and the Depressive Realism Effect. A.P. Field, Learning to Like (or Dislike): Associative Learning of Preferences.

Are preventive and generative causal reasoning symmetrical? Extinction and competition

We tested whether preventive and generative reasoning processes are symmetrical by keeping the training and testing of preventive (inhibitory) and generative (excitatory) causal cues as similar as possible. In Experiment 1, we extinguished excitors and inhibitors in a blocking design, in which each extinguished cause was presented in compound with a novel cause, with the same outcome occurring following the compound and following the novel cause alone. With this novel extinction procedure, the inhibitory cues seemed more likely to lose their properties than the excitatory cues. In Experiment 2, we investigated blocking of excitatory and inhibitory causes and found similar blocking effects. Taken together, these results suggest that acquisition of excitation and inhibition is similar, but that inhibition is more liable to extinguish with our extinction procedure. In addition, we used a variable outcome, and this enabled us to test the predictions of an inferential reasoning account about what happens when the outcome level is at its minimum or maximum (De Houwer, Beckers, & Glautier, 2002). We discuss the predictions of this inferential account, Rescorla and Wagners (1972) model, and a connectionist model—the auto-associator.

Choosing optimal causal backgrounds for causal discovery

In two experiments, we studied the strategies that people use to discover causal relationships. According to inferential approaches to causal discovery, if people attempt to discover the power of a cause, then they should naturally select the most informative and unambiguous context. For generative causes this would be a context with a low base rate of effects generated by other causes and for preventive causes a context with a high base rate. In the following experiments, we used probabilistic and/or deterministic target causes and contexts. In each experiment, participants observed several contexts in which the effect occurred with different probabilities. After this training, the participants were presented with different target causes whose causal status was unknown. In order to discover the influence of each cause, participants were allowed, on each trial, to choose the context in which the cause would be tested. As expected by inferential theories, the participants preferred to test generative causes in low base rate contexts and preventative causes in high base rate contexts. The participants, however, persisted in choosing the less informative contexts on a substantial minority of trials long after they had discovered the power of the cause. We discuss the matching law from operant conditioning as an alternative explanation of the findings.

Maybe this old dinosaur isn't extinct: What does Bayesian modeling add to associationism?

We agree with Jones & Love (JL but we do not believe in the potential of Bayesianism to provide insights into psychological processes. We discuss the advantages of associative explanations over Bayesian approaches to causal induction, and argue that Bayesian models have added little to our understanding of human causal reasoning.