final location of a ball in flight was
impaired when the researchers
blocked out the arm or hand of
the hurling bowler. Novices were
equally bad at predicting where
the ball was headed regardless of
whether they could see the bowler’s arm or hand.
IN
What’s more, anticipation abilities improved among expert batters when they were allowed to
fully swing their bat while making a prediction, compared with
predicting while standing still or
while only completing the lower-body motion of a swing, the Australian team reported last year in Acta
Psychologica. Novices’ predictive
ability did not improve when they
picked up the bat, suggesting that
success in sports is partly dependent on how effectively the brain
couples the body’s perceptive
machinery to its motor processes.
Milton has suggested that athletes in all fast-ball sports — including baseball and tennis — anticipate
where the ball is headed based on
information derived from watching
their opponents’ movements. What
helps separate elite members of
these sports from novices is a superior ability to sort out the relevant
from the irrelevant physical cues.
Novices’ responses (%)
Of course, once the brain gets
the message, the body still has to
react appropriately.
100
90
80
Shot outcome predictions, elite
60
50
40
30
20
10
0
70
426
568
497
639
710 852 1207
1623 781 923
Time (ms)
Shot outcome predictions, novice
with heavy topspin towards Roger
Federer, Federer’s brain computes
bounce heights from previous top-
spin shots to determine how high
the ball will bounce, so he can pre-
pare a swing well before the ball
rebounds off the ground.
Forward models aren’t set
entirely in mental stone, how-
ever — a good thing, since rarely
are multiple scenarios in sport ex-
actly the same. The ball Nadal hits
toward Federer might be slightly
deflated or could glance off the
baseline, causing the topspin
shot to bounce lower than Federer
would have predicted based solely
on previous topspin shots. If
Federer’s forward model didn’t
make use of current sensory infor-
mation to adjust predictions built
on “priors” — the accumulated
knowledge of all the topspin shots
he has seen before — he wouldn’t
be able to react on the fly when
something unexpected happens.
The brain’s predictive machin-
ery is constantly being updated
with new sensory information as
it executes a motion, a feedback
loop that helps the body main-
tain control over its movement,
Todorov says. “Given your goal,
given where you currently are,
the optimal feedback loop posits
the best way to get there,” he says.
Todorov and other scientists
are finding that athletes’ brains
calibrate forward models in a
manner consistent with Bayes-
ian decision theory, a statistical
approach that combines a con-
tinual stream of new information with
previous beliefs. Because there is a level
of uncertainty associated with sensory
input, the brain has to decide whether
it is going to rely more on the new data
(which could be misleading) or on more
credible (albeit potentially outdated)
priors. Elite athletes, who have acquired
more priors through frequent competi-
tion and practice and who have less noise
in their sensory input and motor output,
will have the edge, Todorov suggests.
S.M. AGLIOTI ET AL/NATURE NEUROSCIENCE 2008
426 497 568 639 710 781 852 9231207 1623
MISS 426 497 568 639 710 781 852 92312071623
Correct
Uncertain
Incorrect
100
90
80
60
50
40
30
20
10
0
70
426
Time (ms)
497
568
639
710 852 1207
1623 781 923
Early clue Expert players (top graph) were better than
novices (bottom graph) at predicting the outcome of a
basketball shot from video clips (top). For experts, correct
answers surpassed uncertain ones (yellow line) before the
ball was released. Novices needed more time (orange line).
Correct
Uncertain
Incorrect
A model plan for the future
In the heat of the game, athletes
have to process the sensory data
they’re taking in to automatically
deliver the best motor response. To save
precious time while performing such
calculations, the brain builds a virtual
representation of the world so it can
predict what might happen next, new
research finds. Called “forward models,”
these mental maps allow athletes to preplan “what they want to accomplish and
how they’re going to accomplish it,” says
Emanuel Todorov of the University of
Washington in Seattle.
The brain readies commonly repeated
actions in the motor cortex just like a
torpedo is loaded into a firing bay. But
the response action can’t fire until the
command to act is given by the forward
model. If an athlete’s forward model
is working well, it determines the best
countermove quickly, reducing delays in
the body’s movement.
