Chickens orient using a magnetic compass

Although behavioural experiments show that a wide range of animals use the earth’s magnetic field as a compass for orientation, evidence from conditioning experiments has proved elusive [1]. In birds, the only two successful attempts of operant conditioning to magnetic stimuli [2,3] both involved magnetic anomalies rather than changes in magnetic direction. By using the young domestic chick’s motivation to locate a hidden social stimulus [4], we have demonstrated the first conditioned magnetic compass response in birds and show that the ability to orient using magnetic cues has been retained after thousands of years of domestication [5]. Eight layer-strain domestic chicks were imprinted on a red table tennis ball and subsequently trained to locate this ball hidden behind one of four screens in the north, east, south or west corners of a square arena (Figure 1A). Each chick was released in the centre of the arena to search for the ball that had been hidden behind one of the screens. After finding the ball and remaining with it for one minute (reward), the chick was returned to the home pen while the arena was rotated. The chick was re-introduced after approximately 2–5 minutes into the arena for the next trial to locate its reward. Training continued until the chick approached the ball without deviation in three consecutive training trials (criterion). Unrewarded tests — no ball hidden behind screen — followed, in which we recorded the direction of the first screen that the chick walked behind. Chicks received five tests in each of three conditions: first, the local geomagnetic field (control tests; 56000 nT, –62° inclination); second, an experimental magnetic field with magnetic North shifted by 90° clockwise to geographic East (shifted-north tests); and third, a field with the vertical component inverted, resulting in an inclination of +62° (inclination tests). These tests were presented in a random order and separated by one successful training trial. Training and the 15 tests took 2–4 days for each chick; the chicks were subjected to this procedure twice, at 10–14 and again at 19–23 days post-hatching (hence age was a within-subject factor in the analysis). Chicks reached criterion in 10–22 training trials and required on average 2.0 ± 0.18 training trials between tests during the first testing period, and reached criterion in 5–11 training trials and required on average 1.7 ± 0.16 training trials between tests during the second testing period. The behaviour of the chicks was axial (Table 1), that is, they preferred the correct screen and the opposite screen in the control tests, as well as in the two experimental fields (Figure 1B). In control tests, 76 ± 3.7% (chance = 50%; t = 7.5, P < 0.0001) of their choices lay along this axis. In the shiftednorth tests, 78 ± 3.1% of their choices lay on an axis shifted by 90° (F1,6 = 49.4, P < 0.001). This response to the shift in magnetic north was independent of the age of the chicks (F1,6 = 0.21, P = 0.66) and their sex (F1,6 = 1.1, P = 0.34). In the inclination tests, the choices were as in the control tests, with 79 ± 2.3% of the choices on the same axis (F1,6 = 0.2, P = 0.67). The chicks were not orienting using other non-magnetic cues from inside or outside the arena (see supplemental data); their orientation clearly depended on the direction of the ambient magnetic field.