[This paper passed me by last year, but I think it is worth drawing attention to]
Quite rightly, much attention is given to insects and the ways in which they demonstrate “smartness” in their day-to-day behaviour. Exploring the bounds of these cognitive abilities is, of course, a key part of the field of Comparative Cognition but there are different ways to approach this endeavour. In this review, Doering and Chittka, highlight the problems of a top-down approach to comparative science. For the authors, top-down animal cognition research focuses on cognitive capabilities that have been defined and studied in a human context, then redefines them and seeks them in animals. Examples given here include teaching, culture, consciousness and personality. Problems are generated because of the terminological ambiguity and the focus on ‘clever’ animals. Of course, being able to use similar words to describe behaviours in ants and humans, says nothing about the mechanisms underpinning those behaviours. In contrast, Doering and Chittka propose a bottom-up approach to animal cognition where one’s hypotheses and experiments are inspired by basic behavioural observations. I think I agree.
TF Döring and L Chittka (2011) How human are insects, and does it matter? – Formosan Entomologist, 31: 85-99 pdf here
The level of performance shown by insects in their optic flow mediated behaviours has led many computational scientists to try and develop algorithms that can capture both the performance and the (presumed) computational efficiency of insect optic flow methods. In this paper, the authors develop insect inspired methods for optic-flow based PI. Crucially, the authors have built an extra layer to their algorithm which attempts to mitigate against the cumulative errors found in self-estimates of position.
Here is the abstract: “Some insects use optic flow (OF) to perform their navigational tasks perfectly. Learning from insects’ OF navigation strategies, this article proposes a bio-inspired integrated navigation system based on OF.The integrated navigation system is composed of an OF navigation system (OFNS) and an OF aided navigation system (OFAN). The OFNS uses a simple OF method to measure motion at each step along a path. The position information is then obtained by path integration. However, path integration leads to cumulative position errors which increase rapidly with time. To overcome this problem, the OFAN is employed to assist the OFNS in estimating and correcting these cumulative errors. The OFAN adopts an OF-based Kalman filter (KF) to continuously estimate the position errors. Moreover, based on the OF technique used in the OFNS, we develop a new OF method employed by theOFANto generate the measurement input of the OF-based KF. As a result, both the OFNS and the OFAN in our integrated navigation system are derived from the same OFmethod so that they share input signals and some operations. The proposed integrated navigation system can provide accurate position information without interference from cumulative errors yet doing so with low computational effort. Simulations and comparisons have demonstrated its efficiency.”
Chao Pan, He Deng, Xiao Fang Yin & Jian Guo Liu (2011) An optical flow-based integrated navigation system inspired by insect vision Biol Cybern (2011) 105:239–252
The consensus view within our field is that much of the navigational prowess of social insects is based on the learning of procedural instructions for route guidance. One interesting question involves the extent to which routes can be optimised based on the locations that need to be visited – something akin to the travelling salesman problem. Mathieu Lihoreau and colleagues have been studying these questions with bumblebees learning routes in a small-scale flight chamber. The basic result is that bees will produce near optimal routes to visit a set of feeders. In this pair of papers, Lihoreau et al follow-up on this result by asking whether routes emerge from a nearest neighbour rule (they don’t) and whether the optimisation of routes is based on resource quality as well as distance (it is). 2012 promises to be a fascinating year for this project as at the recent Bielefeld workshop Mathieu previewed results from a large scale version of the experiment – and we all await the articles from these experiments.
Lihoreau M, Chittka L, Le Comber SC, Raine NE. (2011). Bees do not use nearest-neighbour rules for optimization of multi-location routes. Biology Letters.
Lihoreau M, Chittka L, Raine NE. (2011). Trade-off between travel distance and prioritization of high reward sites in traplining bumblebees. Functional Ecology, 25:1284-1292.
Of all the broad family of snaphot-style models of local visual homing, the Average Landmark Vector was always a personal favourite because of its elegant simplicity. The basic procedure is that a panoramic scene can be collapsed to a simple vector which represents the average of unit vectors pointed at all landmarks in the scene. Homing occurs through a comparison of current ALV and an ALV stored at a goal location. Since the initial demonstration of the ALV, various researchers have tried to transfer the elegance of the model to natural complex scenes. Ramisa et al present new research in this direction. For their implementation they use invariant features such as finding local extrema following a difference of gaussians filtering or MSER (Maximally Stable Extremal Regions). In a series of tests both techniques work well and are stable across changes in illumination.
Arnau Ramisa Alex Goldhoorn David Aldavert Ricardo Toledo and Ramon Lopez de Mantaras “Combining Invariant Features and the ALV Homing Method for Autonomous Robot Navigation Based on Panoramas” J Intell Robot Syst (2011) 64:625–649
For insects, we tend to think of visual learning in the context of flower learning or navigation. However, some social insects that live in well lit nests have the opportunity to augment their social skills with visually mediated face recognition of conspecifics. Sheehan and Tibbetts have investigated the face recognition skills of one such species Polistes fuscatus. They conclude that P. fuscatus are not general visual specialists but their superior visual learning, relative to a closely related species P. Metricus, is specific to face recognition. This suggests an independent evolution of a specialised module for recognition of conspecifics.
Sheehan, M. J. & Tibbetts, E. A. Science 334, 1272–1275 (2011).
Certain flowers (in this case Alcea setosa) have a fixed number of nectaries. Thus there is an opportunity for increased foraging efficiency to drive the selection of strategies that correlate with counting – i.e. move on to the next flower after draining 5 nectaries. Bar-Shai et al have studied bumblebees visiting such flowers and show that bee behaviour contains a component which correlates with counting. In this recent paper, they perform the same analysis for solitary bees foraging at the same flowers, however solitary bees don’t show any behavioural component that might be related to counting.
Noam Bar-Shai, Tamar Keasar, Avi Shmida, How do solitary bees forage in patches with a fixed number of food items?, Animal Behaviour, Volume 82, Issue 6, December 2011.
Re-posted. Article now published.
Within the vertebrate navigation literature there is a large volume of research on the use of geometrical cues for orientation within a bounded space. This is part of a language of orientation cues which presupposes that animals segregate the world in cues of different types (features, distal cues, geometry cues etc.). By showing in their 2009 paper that ants makes similar “geometrical” errors to vertebrates, Wystrach et al opened a debate about what types (levels even) of navigation mechanism are required to explain the apparent segregation of the world into geometrical and feature cues. The strong suggestion then was that a simple view based mechanism can account for the the “appearance” of both orientation strategies. Here, the same authors go into a greater level of detail about how view-based strategies can explain the performance of ants in a rectangular arena task. The article has particular value in demonstrating the value of recording in detail the visual information available to an animal during an orientation experiment.
Wystrach, A., Cheng, K., Sosa, S., & Beugnon, G. (2011). Geometry, Features, and Panoramic Views: Ants in Rectangular Arenas. Journal of Experimental Psychology: Animal Behavior Processes.
Advance online publication. doi: 10.1037/a0023886
Given that the constraints and requirements for an ant forager are very different to those for an interior worker, it is no surprise that there are brain changes that go along with the transition from ‘housework’ to foraging. Previous studies have documented changes in the Mushroom Bodies that occur at this transition and there appears to be a strong role for light as a trigger for these changes. In this paper Stieb et al tie together the neuronal and behavioural aspects of the transition to foraging. Behavioural observations of late stage interior ants shows a transition to digging -which exposes ants to light. The authors report that exposure to light increases locomotor activity and needs to be over more than 1 day to instigate the necessary neuronal development. Following the digging, there follows 2 days of short orientation runs which precede real foraging.
The emergent story is of an exquisitely organised set of interactions between behaviour and neuronal development and many interesting questions now arises, such as how these orientation walks are functionally organised and whether there is a developmental role which sets then apart from learning walks.
Visual experience affects both behavioral and neuronal aspects in the individual life history of the desert ant Cataglyphis fortis. Sara Mae Stieb, Anna Hellwig, Rüdiger Wehner and Wolfgang Rössler Developmental Neurobiology (2011) DOI: 10.1002/dneu.20982