MAGAZINE OF THE SOCIE TY FOR SCIENCE & THE PUBLIC MAGAZINE OF THE SOCIETY FOR SCIENCE & THE PUBLIC
deputy;managing;editor,;news Matt Crenson
deputy;managing;editor,;features Elizabeth Quill
contributing;editor Alexandra Witze
astronomy Ron Cowen
behavioral;sciences Bruce Bower
biomedicine Nathan Seppa
chemistry/interdisciplinary;sciences Rachel Ehrenberg
life;sciences Susan Milius
molecular;biology Tina Hesman Saey
editorial;assistant;Camille M. Carlisle
web;specialist/editorial;secretary Gwendolyn K.N. Gillespie
science;writer;intern Daniel Strain
contributing;correspondents Laura Beil, Susan Gaidos,
design;director Beth Rakouskas
assistant;art;directors Theresa Dubé, Erin Feliciano
circulation;manager Tosh Arimura
advertising Jonathan Sismey, Eastern Director;
Mike Walker, Western Director
advertising/circulation;associate Kerwin Wilson
permissions Evora Swoopes
board;of;trustees;chairman;H. Robert Horvitz
vice;chair;Jennifer Yruegas secretary;Gayle Wilson
treasurer;Michela English members;Craig R. Barrett, S. James Gates Jr.,
Tom Leighton, Alan Leshner, Stephanie Pace Marshall, Patrick McGovern, Joe Palca,
Anne C. Petersen, Robert W. Shaw Jr., Frank Wilczek; Elizabeth Marincola, ex officio
executive;assistant Amy Méndez
accounting;manager;Lisa M. Proctor senior;accountant;Sivakami Kumaran
outreach;director;Jennifer Carter development;director BJ Cortis
communications Caitlin Jennings external;affairs;specialist;Nancy Moulding
domestic;fairs;Laurie Demsey volunteers;and;special;awards;Diane Rashid
international;fairs;specialist;Jinny Kim associate;Laura Buitrago
manager,;network;administration;James C. Moore
logistics Anthony Payne facilities;Paul Roger
information;technology;Thomas D. Smith, Amy An, Gregory A. Sprouse
1719 N Street NW, Washington, DC 20036 Phone (202) 785-2255
Subscriptions email@example.com;Editorial/Letters firstname.lastname@example.org
Digital edition provided by Texterity, www.texterity.com
Science News (ISSN 0036-8423) is published biweekly, for $54.50 for 1 year or
$98 for 2 years (international rate $80.50 for 1 year or $161 for 2 years) by Society
for Science & the Public, 1719 N Street N W Washington, D. C. 20036.
Preferred periodicals postage paid at Washington, D.C., and an additional mailing office.
Subscription Department: PO Box 1205, Williamsport, PA 17703-1205. For new
subscriptions and customer service, call 1-800-552-4412.
Postmaster: Send address changes to Science News, PO Box 1205, Williamsport, PA
17703-1205. Two to four weeks’ notice is required. Old and new addresses, including zip
codes, must be provided. Copyright © 2011 by Society for Science & the Public. Title registered as trademark U. S. and Canadian Patent Offices. Printed in U.S.A. on recycled paper.
relies on Bayesian brains
Science and guesswork evolve hand in
hand, along with the human brain itself.
Think about it. Brains evolved to
make successful guesses about the best
survival strategies, helping their bodies
live longer and reproduce more successfully than they otherwise might. At the
same time brains also evolved the abil-
ity to engage in scientific investigation about things pretty
far removed from the perils of the savanna.
Turns out, though, that there is a connection between
these seemingly discordant abilities. Brace yourself: It has
something to do with Bayesian statistics.
In Bayesian statistical analysis, a “prior” probability (an
estimate made before data are collected) is factored into
calculations of the likelihood that a hypothesis is correct.
Bayesian math annoys some traditionalists who cling to
good old frequentist statistics, and the proper use of statistics in science is certainly an enormously complicated issue
(SN: 3/27/10, p. 26). Bayesian math also poses computational
challenges; in complex cases it can be more perplexing than
calculating an NFL quarterback’s efficiency rating.
Nevertheless, neuroscience research suggests that brains
are naturally Bayesian: They operate on the basis of past
experience but update previous beliefs as new evidence
warrants doing so. On Page 13, Laura Sanders discusses one
intriguing instance where the brain bases its judgments on
prior probabilities (perhaps instilled in early evolutionary
times). This example involves the brain’s tendency to rate
an object’s motion as slow. Since most objects don’t move
very fast, when you have to guess about an object’s velocity,
it’s best to go with what’s most probable. If new observations
differ from expectations, your brain should compensate by
modifying its likelihood guesstimates in the future.
Science as a whole proceeds in that very way: Conclusions
based on old evidence are modified as new evidence arrives.
In Bayesian terms, science reflects a cognitive strategy of
basing degree of belief in a hypothesis not just on data collected in an experiment, but also on the probable truth of the
hypothesis as estimated before collecting the new data.
If that’s confusing, you have discerned why Bayesian
methods haven’t been universally adopted in science already,
even if the brain itself uses them instinctively without doing
the math. ( Well, the brain does do the math. It just doesn’t
show its work.) —Tom Siegfried, Editor in Chief
Society for Science & the Public is a 501(c) 3 nonprofit corporation founded in 1921. The vision of Society for Science & the Public is to promote the understanding and appreciation
of science and the vital role it plays in human advancement: to inform, educate, inspire. Visit Society for Science & the Public at www.societyforscience.org. Republication of any portion of
Science News without written permission of the publisher is prohibited. For permission to photocopy articles, contact Copyright Clearance Center at 978-750-8400 (phone) or 978-750-4470
(fax). Advertising appearing in this publication does not constitute endorsement of its content by Science News or Society for Science & the Public.