Cognitive Biases in Human Causal Learning

Abstract

The main aim of this work was to look for cognitive biases in human inference of causal relationships in order to emphasize the psychological processes that modulate causal learning. From the effect of the judgment frequency, this work presents subsequent research on cue competition (overshadowing, blocking, and super-conditioning effects) showing that the strength of prior beliefs and new evidence based upon covariation computation contributes additively to predict causal judgments, whereas the balance between the reliability of both, beliefs and covariation knowledge, modulates their relative weight. New findings also showed "inattentional blindness" for negative or preventative causal relationships but not for positive or generative ones, due to failure in codifying and retrieving the necessary information for its computation. Overall results unveil the need of three hierarchical levels of a whole architecture for human causal learning: the lower one, responsible for codifying the events during the task; the second one, computing the retrieved information; finally, the higher level, integrating this evidence with previous causal knowledge. In summary, whereas current theoretical frameworks on causal inference and decision-making usually focused either on causal beliefs or covariation information, the present work shows how both are required to be able to explain the complexity and flexibility involved in human causal learning. Keywords: causal learning, causal inference, cognitive bias, beliefEl objetivo de este trabajo fue la búsqueda de sesgos cognitivos en la inferencia de relaciones causales para descubrir qué procesos psicológicos modulan el aprendizaje causal. A partir del efecto de la frecuencia de juicio, este trabajo presenta investigación consecuente sobre competición entre claves (ensombrecimiento, bloqueo o súper-condicionamiento) para demostrar cómo la fuerza de las creencias previas y la evidencia sobre la covariación de cada causa contribuyen aditivamente en los juicios causales y en la toma de decisiones, siendo su fuerza relativa modulada por la fiabilidad otorgada a cada tipo de información. Nuevos datos muestran también la incapacidad para detectar relaciones causales incidentales preventivas, pero no generativas. Esta "ceguera inatencional" parece deberse a un fallo en la codificación o recuperación de la información. Todos estos datos revelan que una arquitectura cognitiva del aprendizaje causal debe basarse en tres niveles. El primer nivel sería responsable de la codificación de los eventos en cada ensayo. El segundo nivel computaría la nueva evidencia a partir de la información recibida del primer nivel. En el tercer nivel, el individuo debe interpretar e integrar toda esta información con su conocimiento causal previo. En suma, los modelos sobre juicios de causalidad y toma de decisiones normalmente se han centrado en el efecto exclusivo de las "creencias y conocimiento causal" o de la "experiencia directa y covariación" entre causas y efectos. Este trabajo demuestra que ambos tipos ...

The propositional approach to associative learning as an alternative for association formation models

“…For instance, studies have confirmed that a stimulus is more likely to be seen as a cause of an event if participants have prior knowledge about a mechanism by which the stimulus can cause the event (Bullock, Gelman, & Baillargeon, 1982;Lien & Cheng, 2000). Also, new evidence is interpreted in the light of old evidence for a certain relation (e.g., Maldonado, Catena, Perales, & Cándido, 2007); hence, both abstract and specific prior knowledge determine associative learning in a way consistent with propositional models.…”

CMRules: Mining sequential rules common to several sequences

“…We refer the reader to articles describing CTS [6,30] for full details about its architecture and its learning mechanisms. We have integrated the CMRules algorithm in CTS to give it the capability of finding the cause of learners' mistakes (the ability of finding causes is referred as "causal learning" in the cognitive science literature [31,32,33]). To do so, we have modified CTS so that it records each of its executions (interactions with learners during a training session) as a sequence in a sequence database.…”

Identifying Causes Helps a Tutoring System to Better Adapt to Learners during Training Sessions

“…Furthermore, causal learning is the process through which we come to infer and memorize an event's reasons or causes based on previous beliefs and current experience that either confirm or invalidate previous beliefs [10]. In the context of CELTS, we refer to Causal Learning as the use of inductive reasoning to generalize causal rules from sets of experiences.…”

“…The goal of CELTS' Causal Learning Mechanism (CLM) is two-fold: 1) is to find causal relations between events during training sessions in order to better assist users; and 2) to implement partial procedural learning in CELTS' Behavior Network (BN) 1 , which is based on Maes' Behavior Network [11]. To implement CELTS' CLM, we draw inspiration from Maldonado's work [10] that defines three hierarchical levels of causal learning: 1) the lowest level, responsible for the memorization of task execution; 2) the middle level, responsible for the computation of retrieved information; and 3) the highest level, responsible for the integration of this evidence with previous causal knowledge.…”