1. Insect-level intelligence

2. Information for performing tasks Learning about home – a routine for acquisition Exploration and the return from newly discovered sites Learning routes – scaffolding Multiple routes and memory retrieval Representation of space: routes not maps

3. Behaviours generally evolve within a specific ecological niche and the strategies and underlying neural systems may only operate effectively in that niche.

4. Predator avoidance: categorisation using simple features Zeil and Hemmi, 2006

5. Crabs live on a flat surface where all objects above the horizon are classified as predators Zeil and Hemmi 2006

6. Unexpected sophistication: Crabs at first run to their burrow whenever they spot movement above the horizon. But, if objects are presented repeatedly, their escape attempts habituate, unless the object is a dummy bird. (Hemmi and Zeil)

8. Behaviour is also guided by complex information Burrows are a valued resource and fiddler crabs keep their burrow under constant surveillance. They return rapidly to it if an intruder threatens to take it over. Zeil and Hemmi 2006

9. A crab starts to rush home when an intruder comes within a set distance of the burrow. How does it gauge this distance when the hole is invisible? Crab can know the distance of the intruder from retinal elevation and of the burrow from path integration. These egocentric measures must be combined to give the allocentric distance of intruder from burrow.

10. Ants, social bees and wasps collect food for their nest. Individuals stick to one or few foraging sites and learn fixed routes between these sites and their nest. Cataglyphis bicolor

11. Collett, Dillman, Giger and Wehner, 1992 Homeward routes of Cataglyphis bicolor

12. Behavioural routines for learning A bumblebee leaving its nest for the first time may be gone for an hour and then return to a small nest hole with a load of pollen. On departure. the bee performs an elaborate flight that helps acquire landmark information that can guide its return. The structure of such learning flights reveal efficient strategies for acquiring visual information.

13. Collett (1995) 10 cm Learning flight of Vespula vulgaris leaving feeder 10 cm

14. Two learning flights from one wasp Moments during six of the wasp’s flights when it faces the feeder (+)

15. Return flights to feeder Note consistent body orientation when is close to the goal

16. Body orientation in learning and return flights Wasp 1 Wasp 2 Learning flight (when facing feeder) Return flight (when close to feeder)

17. Wehner, Meier, Zollikofer (2004) Route learning in Cataglyphis starts with exploration. Successive foraging trips of an unsuccessful ant

18. White numbers show successful trips Wehner, Meier, Zollikofer (2004) A luckier ant

19. Wehner 1982, 1990 Path integration encodes the coordinates of an ant’s current position relative to the nest so providing it with the information to go straight home.

20. Ants encode the path integration coordinates of a newly discovered feeding site into long term memory and so can return directly to the site through path integration. Collett, Collett, Wehner, 1999

21. Such straight routes due to path integration are modulated by innate visuo-motor responses to objects.

22. Collett, Collett and Wehner (2000) Route modulation by barrier

23. Graham and Collett (2002)

25. Graham, Fauria, Collett (2003) Route modulation by beacon

26. Scaffolding allows route to be acquired simultaneously along the whole sequence. It also means little chance of learning the wrong thing, so no danger in acquiring routes fast. Why does the ant bother to learn routes if it can reach its destination through path integration?

27. Route memories are linked to motivational state

28. Food-ward and homeward routes overlap Wehner (2003)

29. Formica rufa

30. Harris, Hempel de Ibarra, Graham and Collett (2006)

32. Wehner et al. (2006) Food-ward route not recognised by ant in homeward motivational state

33. Memories primed by panoramic context Collett, Fauria, Baron, Dale 1997

34. Honeybees in a 2 compartment maze in one place Collett, Baron, Sellen,1996

35. Spatial representation Ants use visual landmarks in a procedural rather than a map-like way. They are attracted towards places defined by landmarks or they associate motor commands with landmarks. But landmarks are not labelled with positional coordinates.

36. Independence of PI and landmarks: PI coordinates are not reset by encountering familiar landmarks Collett et al. 2003

37. Routes are independent of PI state: repeated homeward routes (red and black) do not differ from normal trajectories (grey). Kohler and Wehner 2005 But home vector changes. After 1 st repetition of homeward route there is no home vector. After 2 nd repetition, home vector points away from nest. Andell and Wehner, 2005

38. In mammals robust and flexible navigation is ensured by combining path integration and visual information. Separating the two strategies also gives robustness, one strategy can take over when the other fails, and errors in one strategy do not contaminate the operation of the other. A possible hallmark of insect intelligence is that it comprises smart, but specific and independent modules.