|
Basic InformationLatest NewsScientists Create Embryos With Cells From Monkeys, Humans'Game of Thrones' Study Reveals the Power of Fiction on the MindScientists Create Human Tear Glands That Cry in the LabAHA News: How Grief Rewires the Brain and Can Affect Health – and What to Do About ItCould Taking a Swing at Golf Help Parkinson's Patients?Autopsy Study May Explain Why Some COVID Survivors Have 'Brain Fog'Gene Study Probes Origins of Addison's DiseaseCould a Common Prostate Drug Help Prevent Parkinson's?AHA News: Hormones Are Key in Brain Health Differences Between Men and WomenNerve Drug Might Curb Spinal Cord Damage, Mouse Study SuggestsIs There a 'Risk-Taking' Center in the Brain?AHA News: Dr. Dre Recovering From a Brain Aneurysm. What Is That?Can 2 Nutrients Lower Your Risk for Parkinson's?New Clues to How Cancers Originate in the BrainBrain May Age Faster After Spinal Cord InjuryScans Reveal How COVID-19 Can Harm the BrainWhat Loneliness Looks Like in the BrainNeurologists Much Tougher to Find in Rural AmericaCOVID-19 Survival Declines When Brain Affected: StudyAs Testing Costs Rise, Neurology Patients May Skip ScreeningGene Therapy Shows No Long-Term Harm in Animals: StudyCould Gene Therapy Cure Sickle Cell Disease? Two New Studies Raise HopesCocoa Might Give Your Brain a Boost: StudyLockdown Loneliness Could Worsen Parkinson's SymptomsChildhood Lead Exposure Tied to Brain Changes in Middle AgeStaying Social Can Boost Healthy 'Gray Matter' in Aging BrainsDNA Analysis Might Reveal Melanoma RiskGenetics Might Explain Some Cases of Cerebral PalsyDiabetes Drug Metformin May Protect the Aging BrainNew Research Links Another Gene to Alzheimer's RiskYour Sex Affects Your Genes for Body Fat, Cancer, Birth WeightExperimental Drug Shows Promise Against ALSCould Gene Therapy Stem the Damage of Parkinson's?Genetic Research May Help Identify Causes of StillbirthBlood Test Heralds New Era in Alzheimer's DiagnosisMore Clues to the Genes Behind Hearing LossScientists Move Closer to Mapping Entire Human GenomeBlood Test May Reveal Concussion Severity With Accuracy of Spinal TapDeep Brain Stimulation May Slow Parkinson's, Study FindsStroke, Confusion: COVID-19 Often Impacts the Brain, Study ShowsYour Genes May Affect How You'll Heal If WoundedEven Without Concussion, Athletes' Brains Can Change After Head Jolts: StudyHealthDay In-Depth The AI Revolution: For Patients, Promise and Challenges Ahead">HealthDay In-Depth The AI Revolution: For Patients, Promise and Challenges AheadBlood Test Might Predict Worsening MSKeto Diet Might Change Your Gut in More Ways Than OneParkinson's Patient Improving After First-Ever Stem Cell TherapyKey Areas of the Brain Triggered in Recent Heart Attack SurvivorsFirst Good Evidence That Brain Hits 'Replay' While You SleepSome NFL Players May Be Misdiagnosed With Brain Disease: StudyGreenhouse Gases Bad for Your Brain Questions and AnswersLinksBook Reviews |
| |
by J. E. R. Staddon MIT Press, 2001 Review by Helge Malmgren, M.D., Ph.D. on Nov 21st 2002 
This is an important
book which should be of interest to many researchers and advanced students in
learning theory, cognitive science and cognitive philosophy. The authors main
goal is to show the value of learning models which are neither physiological
nor cognitive but rather of a black box character. Such models are framed in
terms of states in a finite or infinite state-space, and describe the
development over time of these states and the interactions between them while
leaving open the the question of their intrinsic nature. (They may for example
be activation states of neural network,
but can also be states of biochemical reaction systems on the cell or sub-cell
level.) This reviewer does not agree with the author in the characterisation of
such models as behavioristic. Abstract explanatory theories of behavior
would be more approriate. This is even more true since, as Staddon himself
acknowledges in the first chapter of the book, the leading behaviorist Skinner
shunned explanatory theories. I mention this point because it exemplifies the
fact (as I see it) that the details of the philosophical and metascientific
arguments in Staddons book are sometimes less convincing than the authors own
substantial contributions to the modelling of adaptive behavior.
However, I
completely agree with the authors main thesis about the necessity of this kind
of abstract theorising, and I find Staddons own attempts in the field
interesting, provoking and not seldom revealing. He tries to show, not only by
detailed theoretical arguments but also by means of fitting model parameters to
a wealth of experimental results, that several seemingly complex and (on the
face of it) cognitive behaviors could indeed be explained by postulating very
simple mechanisms. Of these mechanisms, the leaky integrator (well known from
electrical curcuit theory, and introduced by Staddon in Chapter 4) and cascades
of such integrators play a prominent role. Postulating chains of integrators
with proper parameters, Staddon succeds in modelling not only habituation and similar
simple adaptive patterns (Chapter 6) but also for example (Chapters 7-9) the
effect on eating behavior of different feeding schedules.
Chapters 10-11 deal
with problems in the theory of associative (classical and instrumental)
learning. Except for a thorough discussion of stimulus generalisation, the bulk
of these chapters (to the slight disappointment of the present reviewer) is
mainly devoted to presentations of problems rather than of possible
solutions. Chapters 12 discusses
possible mechanisms for spatial search and argues for a simple diffusion
model which regards route finding as a kind of stimulus generalisation. The
remaining substantial sections of the book deal with knowledge of time and time
intervals. Here, Staddon argues at lentgh against a certain kind of theories
which he designates as pacemaker-accumulator theories, or more intuitively,
theories which postulate an internal clock. Instead he favours an explanation
in terms of multiple integrators with different time scales (an MTS, or
multiple time scale, model), and goes into great detail to show that the main
body of experimental evidence supports the MTS model. I am not in a position to
judge whether this is so, but the authors arguments are consistently logical
and well presented.
I have not yet
referred to Chapters 2-3 of the book. These chapters differ somewhat from the
rest of the book. Staddon here presents a number of "optimality" theories of behavior, discusses the
evidence for and aginst them, and concludes (not very surprisingly) that such
models capture at most a part of animal and human behavior. In the reviewers
opinion, the present book would have been even better if the author had chosen
to publish this stuff elsewhere. The two chapters are only indirectly relevant
to the main theme of the book and they make it less accessible to readers who
do not share Staddons interest in behavioral economics. In other places,
Staddon could have improved the book by leaving out some of the elementary
stuff. But these objections does not change my general judgement that Adaptive Dynamics is a very valuable and
readable - although difficult - work.
© 2002 Helge Malmgren
Helge Malmgren, Professor,
MD, PhD, Dept. of Philosophy, Göteborg University, Sweden. |