and the use of pure and composite measures to reveal them

**Contents, Part 1: TEXT**

**
1. Introduction**

Antecedents

Notation

Twofold partition as an initial step in understanding a complex process

Processes, measures, factors, and selective influence

Statistical issues

Process decomposition versus task comparison

Relevance of brain measurement to process decomposition

Organization of the paper
**
2. Pure measures: Definitions and inferential logic**

Selective influence of factors on processes and their measures

Hypothesis, prediction, and inference for pure measures

Why all four properties are important

3. Composite measures: Definitions and inferential logic

Composite measures based on summation

-- Illustration: The Additive-Factor Method.

Composite measures based on multiplication

Factorial experiments

4. Introduction to three examples of inference based on derived pure measures

5. Isolation of a timing module in the rat (Example 1)

Two kinds of elaboration of the two-module analysis

Comments

6. Parallel neural modules revealed by the lateralized-readiness potential (Example 2)

Comments

7. Separation of sensory and decision processes by signal detection theory (Example 3)

Why has SDT failed in this respect?

An approximation to full modularity of sensory and decision processes when reinforcer ratio is controlled

Comments

8. Introduction to three examples of inference based on direct pure measures

9. Evidence from selective adaptation for modular spatial-frequency analyzers (Example 4)

Comments

10. Evidence from individual neurons for temporally-delimited (serial) neural modules (Example 5)

Estimation of stage durations

Relation of these findings to the demonstration of separate modifiability

Comments

11. Evidence from fMRI for modular neural processes implemented by anatomically delimited processors (Example 6)

Comments

12. Introduction to four examples of inference based on composite measures

13. Evidence from `probability summation' at threshold for modular spatial-frequency analyzers (Example 7)

Justification of a multiplicative combination rule for non-detection probability

Three tests of the joint hypothesis

Comments

14. Amplitude of the event-related potential as a composite measure, with the combination rule known (Example 8)

The additive-amplitude method

Application of the additive-amplitude method to word classification

Comments

15. Multiplicative combination rule for response rate (Example 9)

A plausible systematic deviation from additivity measured by multiplicative interaction of scaled factor levels

Comments

16. Reaction time as a composite measure: Selective effects of sleep deprivation (Example 10)

Comments

17. General Discussion

**Contents, Part 2: APPENDICES, etc.**
**
Appendix A1. Process decomposition versus task comparison**

Introduction

Qualitative task comparison

-- Behavioral studies of memory.

-- Effects of sleep deprivation

-- Task-specific effects of localized brain lesions.

-- Task-specific effects on localized brain activation.

Quantitative task comparison: Subtraction and division methods

-- Donders' subtraction method.

-- Jacoby's division method.

-- Derived pure measures from subtraction and division methods.

Appendix A2. Considerations in the choice of factors

Unitary factors and qualitative process invariance

Manipulated versus selective factors

Inferential logic when hypotheses about modules specify the factors

Appendix A3. Composite measures, combination rules, and stochastic independence

Measures, combination rules, and plausibility

Multiplication as combination rule: Implications for data analysis of the zero-correlation requirement

Stochastic independence of process contributions as further evidence of modularity

Appendix A6. Serial neural modules revealed by the lateralized readiness potential

Appendix A7. Details of the analysis of brightness discrimination by pigeons

Support for the equal-variance Gaussian detection model: Details

Evaluation of an alternative measure of the decision process

Appendix A9. Pure measures: Factorial experiments with multiple-level factors in selective adaptation

Advantages of a factorial design.

Advantages of multiple factor levels.

Appendix A10. Inferring neural processing stages from single-unit recordings

Classification of neurons

Combining activation times over neurons

Effects of stimulus and response factors on stage durations

Appendix A11 fMRI and modular processes: Requirements and statistical issues

A.1 Requirements for a process-decomposition study

A.2 Statistical issues

Appendix A13 Fitting and testing one-channel and two-channel models of detection

Appendix A15 Numerical scaling of factor levels for multiplicative and hybrid combination rules

Appendix A16 Processing stages as modules

Multiplicative combination rule for the proportion of response omissions

What is a `stage'?

Additive effects of factors on mean reaction time: Alternative interpretations