An Analysis of Associative Learning in Drosophila at the Flight Simulator

Abstract

Most natural learning situations are of a complex nature and consist of a tight conjunction of the animal’s behavior (B) with the perceived stimuli. According to the behavior of the animal in response to these stimuli, they are classified as being either biologically neutral (conditioned stimuli, CS) or important (unconditioned stim- uli, US or reinforcer). A typical learning situa- tion is thus identified by a three term contin- gency of B, CS and US. A functional characteri- zation of the single associations during condi- tioning in such a three term contingency has so far hardly been possible. Therefore, the opera- tional distinction between classical conditioning as a behavior-independent learning process (CS- US associations) and operant conditioning as essentially behavior-dependent learning (B-US associations) has proven very valuable. However, most learning experiments described so far have not been successful in fully separating operant from classical conditioning into single- association tasks. The Drosophila flight simula- tor in which the relevant behavior is a single motor variable (yaw torque), allows for the first time to completely separate the operant (B-US, B-CS) and the classical (CS-US) components of a complex learning situation and to examine their interactions. In this thesis the contributions of the single associations (CS-US, B-US and B-CS) to memory formation are studied. Moreover, for the first time a particularly prominent single asso- ciation (CS-US) is characterized extensively in a three term contingency. A yoked control shows that classical (CS-US) pattern learning requires more training than operant pattern learning. Ad- ditionally, it can be demonstrated that an oper- antly trained stimulus can be successfully trans- ferred from the behavior used during training to a new behavior in a subsequent test phase. This result shows unambiguously that during operant conditioning classical (CS-US) associations can be formed. In an extension to this insight, it emerges that such a classical association blocks the formation of an operant association, which would have been formed without the operant control of the learned stimuli. Instead the operant An Analysis of Associative Learning in Drosophila at the Flight Simulator component seems to develop less markedly and is probably merged into a complex three-way association. This three-way association could either be implemented as a sequential B-CS-US or as a hierarchical (B-CS)-US association. The comparison of a simple classical (CS-US) with a composite operant (B, CS and US) learning situation and of a simple operant (B-US) with another composite operant (B, CS and US) learning situation, suggests a hierarchy of pre- dictors of reinforcement. Operant behavior oc- curring during composite operant conditioning is hardly conditioned at all. The associability of classical stimuli that bear no relation to the be- havior of the animal is of an intermediate value, as is operant behavior alone. Stimuli that are controlled by operant behavior accrue associative strength most easily. If several stimuli are avail- able as potential predictors, again the question arises which CS-US associations are formed? A number of different studies in vertebrates yielded amazingly congruent results. These results in- spired to examine and compare the properties of the CS-US association in a complex learning situation at the flight simulator with these verte- brate results. It is shown for the first time that Drosophila can learn compound stimuli and re- call the individual components independently and in similar proportions. The attempt to obtain second-order conditioning with these stimuli, yielded a relatively small effect. In comparison with vertebrate data, blocking and sensory pre- conditioning experiments produced conforming as well as dissenting results. While no blocking could be found, a sound sensory preconditioning effect was obtained. Possible reasons for the failure to find blocking are discussed and further experiments are suggested. The sensory precon- ditioning effect found in this study is revealed using simultaneous stimulus presentation and depends on the amount of preconditioning. It is argued that this effect is a case of ‘incidental learning’, where two stimuli are associated with- out the need of reinforcement. Finally, the impli- cations of the results obtained in this study for the general understanding of memory formation in complex learning situations are discussed.