ctpToMstoch

This C++ code was used to produce the small mutation limit results in the paper T. Lenaerts, M. Saponara, J.M. Pacheco and F.C. Santos (2023) Evolution of a theory of mind. The functions for producing each result are part of the main file, which needs to be edited and recompiled for each result.

Repo Contents
System Requirements
Compilation Guide
Demo Guide
Results
Citation

Repo Contents

All *.CPP and *.HPP files:

rangen.h/rangen.cpp
ctpgame.hpp/ctpgame.cpp
ctpdata.hpp/ctpdata.cpp
strategy.hpp/strategy.cpp
kernel.hpp/kernel.cpp
main.cpp

System Requirements

Hardware Requirements

This code was built on a standard Macbook pro (Monterey 12.6) with

RAM: 8 GB
CPU: 2.4 GHz Quad-Core Intel Core i5

Software Requirements

OS Requirements

The package compilation was tested on OS X (laptop) and Unix (cluster) operating systems.

To compile you will need

GSL (tested for version 2.6 Unix platform and version 2.7 on mac os X)
GCC (version gcc-9.3) on Unix platform and clang on Mac os X.

Compilation Guide

A CMAKE file is provide. Version 3.16.4 was used on the Unix platform. Compile by running first

cmake .

to create the makefiles. Once that works run

make

which should produce an executable ctpstoch.

Demo

When executing ctpstoch without modifications in the main.cpp, the program will produce for an ICG with L=4 the results for β=0.31 and ε=0.19. Changes to the parameters will produce also the L=6 results.

In main.cpp there are 13 parameters that can be set. These are

length The length of the ICG (default value is 4).
first The payoff player 1 gets in the first step (default value is 0.4).
second The payoff player 2 gets in the first step (default value is 0.1).
factor The growth of the resource at each step (default value is 2.0).
levels Reasoning levels of the individuals (default value is 4).
maxlevel Maximum reasoning level in the population (default value is 4).
epsilon Reasoning error for each individual (default value is 0.18).
betas Selection strength in the stochastic evolutionary dynamic (default value is 0.3).
repeats Number of repeats to calculate the fitness of each individual in the Monte Carlo sampling process (default value is 50000).
psize Population size (default value is 500).
mut Mutation probability (default value is 0.0).
scaling Tuning of difference in transitions between reasoning levels (default value is 1.0).
cost Cost associated with each additional reasoning level (default value is 0.0).
approach type of reasoning process; 0)inertia, 1)payoff conditional and 2) unconditional

In main.cpp there are 7 functions that provide the results needed to reproduce the data in the figures:

runAnalytical : this function determines the stationary distribution and other data for a given β and ε (see Figure 2 for instance).
runBeta : this function determines the stationary distribution and other data for a β within a range and a fixed ε.
runEpislon : this function determines the stationary distribution and other data for a given β and a range of ε values.
runBoth : this function determines for a range of β- and ε-values the RMSD with the experimental data and average belief, reasoning, action and misbelief results (see Figure 1A for instance).
runNetwork : this function generates the networks as visualized in Extended Data Figure 5.
runGradient : this function produces the gradient of selection for specific values of β and ε.
RunGradientEpsilon : this function produces multiple gradients of selections for different ε-values.

Output

The raw output files as well as the R files needed to reproduce the figures can be found in the folder results.

The R-scripts can simply be executed in the same folder as the raw data files.

Citation

When using or extending this code, please cite :

T. Lenaerts, M. Saponara, J.M. Pacheco and F.C. Santos (2023) Evolution of a Theory of Mind ... The official version associated with the paper can be found here:

tlenaert/ctpToMstoch