Jumping spiders as a model system for comparative visual cognition

Uncertainty is a common feature of the natural world. Associative learning, which enables animals to form predictive relationships between contingent events, reduces uncertainty and equips animals to respond adaptively. Associative learning is widespread across the animal kingdom, and has been particularly well characterized in a few key invertebrate model taxa. Invertebrates exhibit a broad repertoire of associative learning abilities, have significant molecular overlap with vertebrate systems, and are considerably more amenable to analyses at cellular and behavioural levels. Jumping spiders (Salticidae) - and arachnids as a whole - have been underrepresented in the comparative cognition literature but possess many traits that make them suitable and interesting models. The principle aim of this thesis was to begin to bridge the gap between spiders and traditional invertebrate model taxa. To do so, I developed methods that permitted rigorous and repeatable analysis of cognitive processes, and assessed how salticids integrate different sources of information during learning and meVmory. My research focuses on an Australian salticid, Servaea incana, but the methods and principles developed here should be relevant to other cursorial arachnids or invertebrates in which vision is well developed. I develop a novel method to produce electric shock platforms that is precise, easy to use, and highly repeatable, and isolate the changes in mobility and behaviour that are dependent on shock. Additionally, to overcome the constraints of commonly used playback methods, I develop an immersive, closed-loop virtual reality system tailored for studies of salticid cognition. Using passive and active avoidance conditioning assays coupled with the aversive electric shock stimulus, and a change detection paradigm in virtual reality, I characterize the critical determinants of learning and memory in salticids. These experiments reveal that the ecological relevance, or ‘salience’ of available visual cues, together with hunger level and the temporal nature of the training protocol, play a significant role in determining performance. At the same time, when spiders are provided with a number of cues predicting an aversive event, they learn an operant escape response while ignoring reliable visual information. These results are very surprising given the extent to which vision is known to mediate behaviour and decision making in jumping spiders.