Actu Physiol Scund 1990, 139, 531-534

How do dogs determine the direction of tracks?

J. B. S T E E N and E. W I L S S O N *
Department of Biology, University of Oslo, Norway, and * T h e Swedish Dog Training
Centre, Solleftei, Sweden

STEEN,J. B. & WILSON,E. 1990. How do dogs determine the direction of tracks? Actu
PhysiolScund 139, 531-534. Received 23 January 1990, accepted 27 March 1990. ISSN
0001-6772. Department of Biology, University of Oslo, Norway, and The Swedish Dog
Training Centre, Solleftel, Sweden.

Two professional tracking dogs were brought in at right angles to 50-m-long tracks to
try to identify the clues that the dogs used to determine the direction of the tracks. When
the dogs were tested on tracks laid by a walking man, they took the direction that the
track-maker had moved significantly more often than predicted by random choice
regardless of whether the person had moved forwards or backwards. When tested on
tracks laid by bicycling on grass or asphalt, the choice of direction was not significantly
different from random, even though the dogs were always rewarded when making
correct choices. When four leather strips were tied to the back tyre of the bicycle before
laying the track, the one dog tested took the correct direction sigpificantly more often
than predicted by random choice. When the leather strips were smeared with sausage
before laying the tracks, the dog took the wrong direction more often than predicted by
random choice. However, when rewarded for making correct choices, the dog learned.
to choose the direction in which the bicycle had actually moved. We suggest that our
dogs determined the direction of tracks by comparing the scents of consecutive prints
and following the prints in the direction of increasing intensity.

Key words: dogs, odour, scent, smell, tracking.

T h e ability to track other animals is of vital
importance to most mammalian predators, Not
only must they be able to follow a track, they
must also be able to determine in which direction
the prey has moved. Some trained service dogs
are almost infallible in the art of following human
tracks. T h i s paper attempts to identify the
sensory clues which dogs use in determining the
direction of a track.
There are only two ways in which the direction
of a track can be determined. One is to study
each footprint and detect the position of heel and
toes. T h i s method is used by humans and
perhaps also by other animals whose dominating
sense is vision. T h e other is to compare
consecutive footprints and determine a direc-
tional gradient. Theoretically dogs can study
Correspondence : Johan B. Steen, Division of Gen-
eral Physiology, Department of Biology, University of
Oslo, PO Box 1051 Blindern, 0316 Oslo 3, Norway.
19
footprints by using both scent and vision, but
scent appears to be the most likely source of
information.
Although we have been unable to find any
systematic studies on how dogs determine the
direction of tracks, the problem itself was alluded
to almost 800 years ago. T h e famous Icelandic
narrator, Snorre Sturlason (122&1235), des-
cribed how two Norwegians, imprisoned by the
Swedes in 1026, fooled their pursuers by tying
reindeer hooves back to front under their boots.
T h e Swedish tracking dogs, it is told, followed
their tracks in the direction indicated by the
footprints of the reindeer and thus ended u p at
the empty prison (a deep hole in the pigsty).
While more recent reports indicate how
accurately dogs are able to determine the
direction of tracks (see Mackenzie & Schultz
1987), no one since Snorre Sturlason seems to
have suggested what clues they may use to do so.
T h e experiments to be described below were
53 1
A C T 139
532 3’.B. Steen and E. Wilsson
\
G&
,;
Fig. 1. Which direction to the prey? A hunting dog’s
never-ceasing decision problem. A drawing by Kai
Fjell.
designed to reintroduce the problem to modern
biologists and suggest a solution.
M A T E R I A L S A N D METHODS
The work was carried out at The Swedish Dog
Training Centre, Solleftei, from 1985 to 1988. After
screening a number of dogs, we’selected two German
Shepherds. These showed superb abilities in choosing
the correct direction of tracks and had the mental
stamina to endure repeated testing. The female
‘Lanja’ was used throughout the study, the male
‘1,affe’ became ill and was only used the first year.
Tracks were laid on grass fields or on a tarred
airstrip. Human tracks were laid by one of the authors
walking on grass in rubber boots or on asphalt in
leather shoes. Tracks were also laid by one of the
authors riding a bicycle on either grass or asphalt.
When making bicycle tracks, care was taken not to
touch the ground with the feet. Each track was at least
50 m long and marked with a pole at both ends. The
dogs were tested on up to 20 tracks a day.
The dogs were always handled by their trainers.
The handler and dog were not present when the tracks
were laid and no one who was present when the results
of the tests were recorded knew the direction of the
tracks. The dogs were tested 20-30 min after the track
had been set. With the dog on a leash, the handler was
told where to go and asked to stop when he knew,
from the position of the poles, that he was about 5 m
from where the middle of the track crossed at right
angles to his direction of approach. The dog was then
unleashed and given the order ‘track’ (in Swedish
‘Spw ’).
The performance of the dogs was scored as correct
when the dog tracked in the direction the track-maker
had moved and as wrong when it tracked in the
opposite direction. The dog was always rewarded
when it made a correct choice, either with a titbit or
with a wooden stick which had been left there by the
track-maker but hidden from the view of both dog and
handler. The dog was not rewarded when it made the
wrong choice. Deviation from random choice (50 : 50)
was tested by a simple binomial test.
RESULTS
I n none of the more than 200 tests did the dogs
fail to find and follow the track. However, they
did not always pick the right direction. Once
they came to the track the dogs would normally
hesitate for a few seconds, often walking back
and forth along the track, before they decided
whether to go left or right. Only in a few cases
did the dog change its direction once it had
followed the track more than 5 m.
Test 1 was designed to establish the dogs’
ability to determine the direction of normal
human tracks laid at a rate of about one step per
second. Laffe tracked correctly in all his four
trials, whereas Lanja had 22 correct versus four
wrong when tracks were laid across a grass field.
When tested with similar tracks on asphalt,
Lanja had 10 correct versus one wrong (Table
1).
Test 2 was designed to test if the dogs
determined the direction of tracks by following
them in the direction from heel to toe of each
footprint. T o this end tracks were laid as in test
1, except that the track-maker walked backwards.
When tested on such tracks, Laffe followed the
direction the track-maker had walked in all his
four tests, while Lanja had nine correct versus
two wrong choices (Table 1). T h i s suggested
that the dogs did not determine the direction of
the tracks by following them from heel to toe.
Test 3 was designed to test if the dogs could
determine the direction of a continuous track.
I n pilot experiments we showed that both dogs
had difficulties in deciding the direction of tracks
laid by dragging a pair of shoes across the field
or by walking with such short steps that the heel
hit the ground at the toeprint of the previous
step.
 Directional tracking in dogs 533

Table 1. The ability of the two dogs to determine the direction of different types of tracks

Dog Track Ground Correct Wrong Signif.

Laffe Normal foot Grass 4 0 -

Lanja Normal foot Grass 22 4 < 0.01

Lanja Normal foot Asphalt 10 1 < 0.05
Laffe Backwards foot Grass 4 0 ~

Lanja Backwards foot Grass 9 2 < 0.05

Laffe Bicycle Grass 10 9 n.s.
Lanja Bicycle Grass 10 11 n.s.
1,anja Bicycle +leather Asphalt 19 3 < 0.01
Lanja Bicycle + leather + hand Asphalt 12 11 n.s.
Lanja +
Bicycle leather +sausage Asphalt
First day 4 12 < 0.05
Second day 10 2 < 0.01

First we tested if the dogs could determine the

direction of continuous tracks laid by bicycling
across a grass field. T h e y had no difficulty in
finding and following the tracks but they were
unable to determine their direction (Table 1).
Despite the fact that they were given the
conventional reward every time they made a
correct choice, their performance did not im-
prove as testing went on. W e therefore concluded
that it was beyond the dogs’ capacity to determine
the direction of continuous tracks.
Test 4 was designed to test if Lanja could
determine the direction of a continuous track if
a point track was superimposed onto it. To test
this we tied four strips of leather, at equal
distances apart, around the rear tyre of the
bicycle. Of 22 such tracks laid on asphalt, Lanja
tracked correctly in 19 cases (Table 1). Evidently
the difference in scent strength from consecutive
leather prints was sufficient to tell the dog in
which direction the bicycle had moved. W e
suggest that she was able to determine a gradient
along the track, the leather scent from the first-
laid contacts with the asphalt being weaker than
the more recent ones.
Test 5 was designed to test the suggestion
from test 4. T o this end we tried to make tracks
where the prints supposedly smelled more at the
beginning of the track than at the end. T h i s was
achieved by applying a scent to the leather strips
which would gradually wear off as the bicycle
touched the ground.
First wc applied smell by rubbing the leather
with the hand before laying the track. O f 23 such
tracks, Lanja took the correct direction in 12
cases but did not improve her performance with
1-16 17-28
Track number.
First day Second day
Fig. 2. Results from an experiment where a dog was
led perpendicularly on to a track laid on asphalt by a
bicycle that had four leather strips smeared with
sausage fastened to the rear tyre. Tracks 1-16 were
laid the first day, tracks 17-28 the second. Correct
direction means the direction in which the bicycle had
moved. We suggest that the dog went mostly in the
wrong direction the first day because she followed the
tracks towards stronger smell. However, since she was
rewarded when taking the opposite direction, she
learned to follow the track in that direction. The dog’s
choice of direction is significantly different from
random on both days (P < 0.05). .,correct direction;
B, incorrect direction.
time even though she was rewarded every time
she did it right.
W e then modified the experiment by smearing
the leather with a sausage. T h e trials were
carried out during two consecutive days, 16 the
first day and 12 the next. During the first day
19-2
534 J. B. Steen and E. Wilsson
Lanja went mostly in the wrong direction, i.e.
towards where the bicycle had started (Fig. 2).
However, during the second day she decided
correctly in 10 out of 12 trials. During the first
day her correct choices came in trial 2, 4, 10 and
14, during the second day the first two were
wrong and the rest correct.
DISCUSSION
Our dogs had little difficulty in determining the
direction in which tracks were laid by regular
walking. They were not fooled by walking
backwards, but consistently followed the di-
rection the track-maker had walked. This shows
that these dogs did not determine the track
direction as the direction from heel to toe in
each footprint, as implied in Snorre Sturlason’s
(cn. 1220-1235) narrative.
However, determining the direction of a
continuous track, whether laid on grass or on
asphalt, was evidently beyond the dogs’ capacity.
This suggests that they determined the direction
in which the track-maker had moved by com-
paring sensory inputs from consecutive prints
and tracking in the direction of recency, i.e.
towards increasing strength of a time-decaying
clue. We propose that scent was most likely the
clue, since it is hard to believe that other sensory
clues were available from tracks on dry asphalt.
Our finding that it was more difficult for our
dogs to determine the .direction of continuous
than of interrupted tracks confirms the general
experience that it is easier to distinguish sensory
stimuli when they were discrete rather than
continuous.
If we assume that each footprint smelled the
same at the moment it was set, and that the scent
evaporated at a constant rate, we can get an idea
of the sensitivity of the dogs’ sense of smell. We
walked at a rate of one step per second and the
tracks were 30 min (1800 s) old when the dogs
were tested. The smell from one print should
therefore theoretically be 1/ 1800 stronger than
that of the foregoing. This indicates that the
dogs were highly sensitive to an odour difference
of this magnitude. Johnson (1977) found that his
dogs made few errors in determining the
direction of tracks if they were less than 30 min
old, whereas they were unable to determine the
direction of 3-h-old tracks. This indicates that,
beyond a certain point, the tracks have aged so
much that the dogs are unable to discriminate
odour strength between prints.
The finding that neither of the dogs improved
its ability to determine the direction of bicycle
tracks despite being rewarded when correct
choices were made suggests that such tracks did
not contain the appropriate sensory clues. On
the other hand, tracks where the scent was
applied in patches, either from a foot or from a
leather strip, appeared to contain such infor-
mation. Lanja back-tracked consistently when
we applied a second scent which presumably
wore off with each contact with the ground. This
supports our suggestion that she determined
direction by tracking towards ever stronger
smelling tracks. However, since she was trained
to go in the direction where she found the
reward, she learned to track in the direction
in which the bicycle had moved. This further
supports the idea that she determined direction
by comparing the strength of consecutive prints.
A practical implication of this study is that
simply following a track and determining its
direction appear to be two separate issues for a
dog. A dog trained to follow a track does not
necessarily learn to determine the track’s di-
rection. In order to teach the dog to take the
direction of the track-maker, the dog’s attention
needs to be focused on that task, for example by
being rewarded when it does it correctly.
It is a special pleasure to acknowledge Professor
Ellisiv Steen, who brought the historical aspects of
this study to our attention. We are also indepted to the
dog handlers Erik Sundin and Barbro Borjesson and
to Professor Holger Ursin and Dr Kathryn T. Mearns
for valuable comments on the manuscript. This study
was supported by the Nordic Council for Game
Research and the Carl Trygger’s foundation.
REFERENCES
JOHNSON, G.R. 1977. Training Dogs-Theory and
Methods. Arner Publ., New York.
MACKENZIE, H.A. & SCHULTZ, J.A. 1987. Frequency
of backtracking in the tracking dog. Appl Anim
Behav S c i 17, 353-359.
STURLASBN, s. (ca. 1220-1235) olav den Helliges
Saga. Verse 141. Recent editions in Norwegian and
English are available.