Asymmetrical number-space mapping in the avian brain

Abstract

When trained to peck a selected position in a sagittally-oriented series of identical food containers, and then required to generalize to an identical series rotated by 90°, chicks identify as correct only the target position from the left end, while choosing the right one at chance. Here we show that when accustomed to systematic changes in inter-elements distances during training or faced with similar spatial changes at test, chicks identify as correct both the target positions from left and right ends. However, ordinal position is spontaneously encoded even when inter-element distances are kept fixed during training (in spite of the fact that distances between elements suffice for target identification without any numerical computation). We explain these findings in terms of intra-hemispheric coupling of bilateral numerical (ordinal) representation and unilateral (right hemispheric) spatial representation of the number line, producing differential allocation of attention in the left and right visual hemifields.

Introduction

An important aspect of numerical cognition is the ability to represent ordinal (serial) relationships, which refers to the ability to identify an object on the exclusive basis of its position in a series of identical objects. Rats are capable of learning to enter a target tunnel solely on the basis of its ordinal position in an array of 6 (Davis & Bradford, 1986) or 18 (Suzuki & Kobayashi, 2000) tunnels. Honey bees are able to find a food source located between the 3rd and the 4th position along a series of 4 identical, equally spaced landmarks (Chittka & Geiger, 1995), they can also identify the 4th position in a series of 5, and generalize it to a novel series of objects (Dake & Srinivasan, 2008).

Young domestic chicks (Gallus gallus) also show sophisticated serial ordering abilities (for a review see Vallortigara, Regolin, Chiandetti, & Rugani, 2010). Week old chicks trained to peck either at the 3rd, 4th, or 6th position in a series of 10 identically spaced locations, sagittally aligned in front of the birds’ starting point, learned accurately to identify such positions (Rugani, Regolin, & Vallortigara, 2007). Control experiments ruled out the possibility that chicks were using distance as a cue. A serendipitous bias for the left hemispace was, however, found in one of the experiments controlling for spatial cues. Chicks, once trained to peck a target position in a sagittally-oriented series of identical elements, were then required to generalize to an identical series, but rotated by 90°. Following rotation, the series was oriented fronto-parallel with respect to the chick, hence the correct position could not be located on the basis of absolute distances from the starting point. Moreover, as both ends were at a same distance from the birds’ starting point, the correct position could be identified either from the left or from the right end. Interestingly, chicks identified as correct only the 4th position from the left end, and not the 4th from the right end, which was chosen at chance.

In subsequent work (Rugani, Kelly, Szelest, Regolin, & Vallortigara, 2010), the same behaviour was also described in a different species of birds, i.e. adult nutcrackers (Nucifraga columbiana), as well as domestic chicks, using a larger series, which comprised 16 (rather than 10) identical and aligned positions. When, at test, the series was rotated by 90° as compared to training, lying fronto-parallel to the bird’s starting position, both species showed a bias for identifying selectively the correct position (either the 4th or the 6th element for separate groups of subjects) from the left but not from the right end.

The preference for targets located on the left hemispace may be due to a bias in the allocation of attention, somewhat similar to that shown by humans and dubbed as “pseudoneglect” (Albert, 1973, Jewell and McCourt, 2000). In fact, “pseudoneglect” phenomena have been described in birds with selective allocation of attention to the left hemifield during free foraging (Diekamp, Regolin, Vallortigara, & Güntürkün, 2005) and in bisection tasks (Diekamp, Manns, et al., 2005, Regolin, 2006). Somewhat similar phenomena of pseudoneglect favouring the left hemifield have been described also for amphibians (Vallortigara, Rogers, Bisazza, Lippolis, & Robins, 1998; and see for general reviews MacNeilage et al., 2009, Vallortigara and Rogers, 2005).

The birds’ bias is also reminiscent of the human mental number line phenomenon. As early as 1880, Galton showed that humans describe and think of numbers as being represented on a mental number line, which is usually oriented from left to right. Modern research provides evidence that number magnitude may be represented on a left-to-right oriented number line (Dehaene, 1997). It remains to be shown, however, whether the spatial orientation of the human mental number line is entirely acquired culturally (i.e. it may be linked to writing and reading rules) or if it depends, at least in part, on biologically specific biases in the allocation of attention in extra-corporeal space. Typically human studies on the mental number line are associated with aspects of number cognition which rely on the ability to represent ordinal relations, i.e. mastering the rule that when one element is added to a given set, the new set becomes larger than the previous one and smaller than the next (see for instance the studies on the so-called SNARC effect, see Dehaene, Bossini, & Giraux, 1993). A simpler and more basic ability is that of identifying an object on the exclusive basis of its position in a series of identical objects, without any association with a magnitude of the distances among ordinal positions. Such an ability may provide a viable strategy for investigating biases in space-number mapping in non-linguistic and non-cultural creatures.

In the present series of experiments we aimed to investigate more precisely the nature of the left-hemi-space bias exhibited by birds during serial ordering tasks. In the experiments by Rugani et al. (2010) birds performed successfully at test, when some of the non-numerical cues available during training, such as distance from the starting point and walking time, could no longer be of any use to solve the task. However, birds could still rely on the distance of the correct position from the beginning of the series as a spatial cue to identify the target. The first experiment was aimed to investigate this issue, by disrupting this source of spatial information, manipulating the distance between the positions in the series, both during training and during the test.