Methodological Problems in Operant Learning Research with Human Infants

Abstract

A great number of the recent studies on infant operant learning use the mobile conjugate reinforcement paradigm with 2‐ and 3‐month‐old infants. This paradigm has proved to be adequate for the study of learning and the transfer of learning. However, difficulties linked to the high attrition rates of subjects, and the variability in performance are frequently observed in studies using other paradigms or in studies with older infants (4–5‐month‐olds). The problems noted in a group of studies seem to be related to important methodological issues such as the brevity of the learning sessions, the choice of dependent measures, and the management of variability in performance. The use of single‐subject designs and the use of variables encountered in infants' daily learning experiences may be the solution to some of the problems.

Detection of sensorimotor contingencies in infants before the age of 1 year: A comprehensive review.

“…At the time scale of one particular playing episode, the infant’s interest for contingencies might be organized to optimize its gain in knowledge (e.g., Kidd et al, 2012; Sim & Xu, 2017; Stahl & Feigenson, 2015). At both time scales, infants’ exploration of the environment is also guided by “scaffolding” through the baby’s social partners (Malcuit, Pomerleau, & Lamarre, 1988; Pomerleau et al, 1992). Caregivers organize the baby’s exploration space so that learning situations are in accordance with the baby’s current abilities and interests.…”

“…The long-term retention of a contingency is positively correlated with the duration of exposure to it and is enhanced by exposure spread over several days Time window between two repetitions of a contingency Watson (1967) 3 months Contingency between gaze location and audiovisual feedback In order to learn a contingency, infants need a time window of 3 s to 5 s between two repetitions of the contingency Millar (1972) 4 months Contingency between hand-pulling and audiovisual feedback The time window between two repetitions of a contingency may be critical only at the beginning of contingency learning Probability of occurrence Watson (1985) 4 months Contingency between kicks and social audiovisual feedback Learning is higher for an "imperfect" than a "perfect" contingency Pomerleau, Malcuit, Chamberland, Laurendeau, and Lamarre (1992) 5 months Contingency between touch of an object and movements of the object…”

“…Moreover, this criterion considers that contingency detection necessarily results in a repetition over a long period of time of the contingent action, a behavior that we (Jacquey, 2019) and other authors (Millar, 1972) seem not to have systematically observed. Although such issues related to the analysis of contingency detection are important (see, e.g., Pomerleau, Malcuit, Chamberland, Laurendeau, & Lamarre, 1992), our aim in this review is more generally to understand the parameters that influence babies’ ability to detect a contingency. Indeed, even though there is an abundance of published studies using operant conditioning, there is to our knowledge no article systematically reviewing the characteristics of baby’s sensitivity to nonsocial sensorimotor contingencies (i.e., contingencies involving the body itself or the external physical world but not social partners).…”

Perception of audiovisual infant directed speech

“…Infant studies stand the risk of attrition, resulting in low external validity. Several factors can decrease this risk (Pomerleau, Malcuit, Chamberland, Laurendeau & et al, 1992) and the less demanding the experimental task requiring sustained attention, the more likely infants will finish the procedure (Oates, 1998). Hence, a relatively simple task may lead to low attrition.…”

Infant responding compared under conjugate- and continuous-reinforcement schedules.

“…Because infant attention span is fluctuating and short, experimental sessions with them are usually brief in order to reduce subject attrition (Hulsebus 1973). While brevity of experimental sessions insures that the infant is kept alert and interested, it may prevent experimenters from measuring fully the impact of the independent variable on dependent measures (Pomerleau & Malcuit, 1992). On the other hand, an operant task, such as the synchronized reinforcement procedure, appears to be more amenable in maintaining responding on the part of the participants being tested.…”