Skip to content

Replicate the neural network model in Cohen et al. (1990). On the control of automatic processes: a parallel distributed processing account of the Stroop effect.

License

Notifications You must be signed in to change notification settings

qihongl/stroop-cohen-etal-1990

Repository files navigation

the stroop model

this repo replicates cohen et al (1990) w/ psyneulink composition (tested on version==0.5.2.1), this is developed for the course NEU|PSY330 Computational Modeling of Psychological Function, Spring 2019

doc

here're some slides i used for the lab

  • stroop_model.py: the definition of the stroop model
  • stroop_stimulus.py: the definition of the stroop task; I also uploaded this helper script this as a package, so you can pip install stroop. Here's its repo
  • run_exp_*.py: run some experiment, where * is the name of the experiemnt
  • show_*.py: analyze and plot the data
  • stroop-feedforward.ipynb and stroop-linear.ipynb: two simplifications of the full stroop model, for teaching purpose

the model

here's the architecture...

model

results

here's the main result, which qualitatively replicate fig 5 from cohen et al (1990). in general...

  • color naming is slower than word reading
  • color doesn't affect word reading but word affect color naming a lot. for example, color naming red green as "red" is much slower than word reading red green as "green", where red green = the word green painted in red.
  • magnitude(interference) > magnitude(facilitation)

rt

here's another way to plot the data. these are kernel density estimates of the reaction time distributions.

rt kde

here's the effect of demand

  • the left panel corresponds to fig 13 a in Cohen et al. 1990
  • note that the right panel is not plotting fig 13 b in the paper -- we don't have a shape naming condition here

demand

and here's the SOA effect - figure 7 in Cohen et al 1990:

soa

decision energy, see Botvinick et al. 2001 and Shenhav et al. 2013:

dec_eng

References:

[1] Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). On the control of automatic processes: a parallel distributed processing account of the Stroop effect. Psychological Review, 97(3), 332–361. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/2200075

[2] Shenhav, A., Botvinick, M. M., & Cohen, J. D. (2013). The expected value of control: an integrative theory of anterior cingulate cortex function. Neuron, 79(2), 217–240. https://doi.org/10.1016/j.neuron.2013.07.007

[3] Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S., & Cohen, J. D. (2001). Conflict monitoring and cognitive control. Psychological Review, 108(3), 624–652. Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/11488380

[4] the code is based on all stroop model scripts I can find from psyneulink example scripts here, and most importantly, this and this

About

Replicate the neural network model in Cohen et al. (1990). On the control of automatic processes: a parallel distributed processing account of the Stroop effect.

Topics

Resources

License

Stars

Watchers

Forks

Packages