Aging and Rule Learning
Gérard Chasseigne, & Peggy Lafon
We have been pursuing the research program initiated in 1997, aimed at examining the effect of age on learning in complex situations. The present work concerns aging and rule learning. It is aimed at examining the effect of age on the ability to learn multiplicative combination rules. In such rules, cue levels do more than simply add their effects to determine the criterion values; the combination rule is a multiplicative one. To our knowledge, there has been no work on such a topic.
The specific situation studied was the learning of the multiplicative relationship between daily tobacco intake, daily alcohol intake and risk of cancer. This situation was chosen as a function of three criteria. Firstly, we wanted to use a concrete situation, that is a situation elderly participants could consider as relevant to everyday life: tobacco smoking, alcohol drinking, and cancer suffering are life events. Secondly, we wanted to use a situation for which a multiplicative combination model existed: A multiplicative type model of the relationship between cigarette dose, alcohol dose and risk of cancer of the esophagus has been proposed (Tuyns, Péquignot & Jensen, 1977). Thirdly, we wanted to use a situation for which it had already been demonstrated that the combination rule implemented by naive participants was not multiplicative: It has been repeatedly demonstrated that naive participants eschew to apply a multiplicative rule when asked to estimate the risk of cancer associated with given alcohol-tobacco consumption combinations. They uniformly apply a disjunctive rule (Hermand, Mullet & Coutelle, 1995; Hermand, Mullet & Lavieville, 1997; Hermand, Mullet, Sorum & Tillard, 2000). That is, they consider that indulging in only one of these two behaviors represents a maximum health risk.
Our overall hypothesis was that elderly people, more than young people or mid-adults would experience difficulties in learning to apply a multiplicative rule. This hypothesis was based on the proposition made by Chasseigne, Mullet and Stewart (1997), and Chasseigne, Grau, Mullet and Cama (1999) that the differences between young and elderly people in probabilistic function learning are mainly related to flexibility of functioning (changing from one default combination hypothesis, here the disjunctive rule, to another combination hypothesis, here the multiplicative rule). Specifically, we expect that (a) young people would pass gradually, through learning, from the use of a disjunctive rule to the use of a multiplicative rule, and (b) elderly people would keep using the disjunctive rule, despite feed-back.
A total of 86 individuals (33 males and 53 females) participated in this experiment (23 young adults aged 18-25, 22 employed persons aged 40-50, 20 retired persons aged 65-74, and 21 older persons aged 75-90). None was institutionalized.
The materials consisted of one set of 25 cards (21 x 3.5 cm), each showing two cue values in the form of tobacco consumption levels (0 pack, 1/2 pack, 1 pack, 1 and 1/2 packs, and 2 packs a day), and alcohol consumption levels (0 glasses, 2 glasses, 1 bottle, 1 and 1/2 bottles, and 2 bottles a day). The criterion value (1-100), which expressed the risk level associated with the tobacco and alcohol consumption levels, was written at the back of each card. This value was computed from Tuyns, Péquinot and Jensen's (1977) rule.
The subjects were told that their task was to forecast the risk of esophageal cancer associated with the various consumption levels. They were provided with the 25 different vignettes. The actual value of the criterion was displayed on the back of the same card as outcome feedback (OFB). Subjects were asked to learn the relationships between the levels of the two indicators and the overall risk level. The subjects were shown six blocks of 25 trials. The first block was a familiarization block without OFB. The second, fourth, and last blocks served as policy capturing blocks (no OFB). The third and fifth blocks were learning blocks in which OFB was provided to the subjects. The experiment was self-paced, and subjects completed the task individually. They took about 1-1.5 hours to complete the experiment.
Our hypotheses were well supported by the data. Before receiving feedback, the participants implemented a disjunctive rule. After receiving a limited amount of feedback (first learning block), these participants had already learned to use a multiplicative rule. Even after receiving a massive amount of feedback (two learning blocks), however, the elderly people still showed difficulties using the multiplicative rule. These results strengthen the proposition put forward by Chasseigne, Mullet and Stewart (1997) and Chasseigne, Grau, Mullet, and Cama (1999) that the differences between young and elderly people in function learning are mainly related to flexibility of functioning (changing one hypothesis to another).
Contact Gérard Chasseigne