(Downloading may take up to 30 seconds.
If the slide opens in your browser, select File -> Save As to save it.)
Click on image to view larger version.

Fig. 2. Suitability of the responses analyzed in this paper. (A-C) Evidence that
the responses were directed at the predicted later point of impact but were
not based on an approximate mechanism. (A) Illustration of the errors
analyzed. A shooter has dislodged a fly, which falls towards its later point
of impact P. At the instant shown, a bystander has just finished its initial
turn and starts to head off with a bearing indicated by the solid line. The
actual position of the fly is at P'. The solid line crosses line of prey
movement (from P' to P) at point S. Errors
and
'
denote the angular deviations of the bystander's bearing with respect to
either the later point of impact P or the actual position P' of the fly
(broken lines). Error
(
') is positive when the respective
reference point P (P') is before S, otherwise the error is negative: in
the situation shown
' is positive and
is negative. (B)
Bearing errors
made with respect to the later point of impact are
symmetrically distributed around zero. (C) In contrast, the distribution of
errors
' with respect to the fly's actual position, is
systematically offset towards positive values (P<0.001,
t-test). Errors
and
' are sampled in intervals of
2°. (D) Distribution of distances the responding fish had to cover from
their initial position towards the later point of impact. Bin width 50 mm. (E)
The time
that remained till prey impact when the fish had sampled the
necessary information about target motion, had finished their turn and were
ready to take off. The histogram shows how remaining times
, binned in
0.02 s intervals, were distributed in the set of responses. (F) The fish
responded over a remarkably wide range of target-flight directions with
respect to their initial orientation. The histogram shows the correspondingly
wide distribution of turn sizes the responding fish made. Bin width is
20°. All histograms are based on the same N=90 responses and are
normalized so that their total frequency equals 1, ticks on ordinate indicate
10% frequency.