Reading focus
Syllabus: How Does Evolution Help Explain Behavior?
Translate the alternative trees into claims about how much independent evidence the observations contain.
Why you’re reading it. Prepare for the phylogenetic regression in MacLean's study while keeping it distinct from Felsenstein's particular contrast procedure.
A question to consider. If ten closely related species share both a broad diet and high task scores, what additional species or comparisons would strengthen the inference?
Why this reading
A comparison of many species can contain fewer independent evolutionary changes than its sample size suggests. Felsenstein shows how shared ancestry undermines ordinary statistical assumptions and develops independent contrasts. The paper connects a statistical comparison to an explicit model of evolutionary history.
Before you read
A phylogeny is a branching account of ancestry. Related species can resemble one another because they inherited features from a common ancestor. An independent contrast compares changes across branches rather than treating every species value as a separate replicate. Brownian motion is a model in which change accumulates with branch length; its use is an assumption, not a claim that evolution lacks selection.
Read for
- Use the alternative trees in Figures 2 and 3 to explain why the same number of species need not supply the same amount of independent information.
- Trace the reasoning behind standardizing contrasts, then examine the difficulties concerning tree reconstruction, branch lengths, and the model of character change. Understanding those assumptions does not require reproducing the derivations.
Consider afterwards
- Two behavioral traits correlate across species, but their differences largely separate two families. What new sampling would make the comparison more informative?
- Why would accounting for ancestry still leave causal direction and measurement unresolved?
Editorial context
This is a methodological foundation with explicitly stated limitations, not a current software manual. Its contrasts and later phylogenetic regression approaches share a concern with dependence but are not interchangeable procedures in every setting. Correcting an uncertainty estimate cannot by itself identify selection, eliminate ecological confounding, or establish that two tasks measure comparable capacities.