Computational Models of Aphasia

Welcome to Computational Models of Aphasia at the University of California, Irvine's Auditory & Language Neuroscience Lab!
This site draws much of its inspiration from the Aphasia Modeling Project.

This website provides statistical methods for applying the two-step, interactive model of lexical access to picture naming responses (Foygel & Dell, 2000).
Users may enter naming response frequencies to obtain Bayesian point and interval estimates of the S-weight and P-weight parameter values with the best fit, or users may enter S-weight and P-weight parameter values to obtain naming response probabilities.

Fit Data
Run Model
References and Contacts

Fit Data

This website uses a Metropolis-Hastings sampling algorithm to approximate the posterior distribution of weight parameters in a lexical network given data.
Weight parameters are assumed to lie on the interval [0, 0.04].

The sampling procedure can take up to 30 seconds to fit the data.

Please enter the observed counts of each response type:

Correct	Semantic	Formal	Mixed	Unrelated	Nonword

Fitting Parameters		Other Parameters
Random Seed			Steps
Jump Sigma	.05		Decay
Chains	2		Intrinsic Noise
Samples	50		Activation Noise
Burnin	20		Initial Activation
			Jolt Activation
		20%	Mixed Error Opportunities

Run Model

This website uses numerical approximation of integrals to calculate response probabilities given lexical network parameters.

Please modify the parameters you wish, and select Run!
Note: Weight parameter values of 0.04 simulate the "normal" or "healthy" lexical network, and lower values simulate damage.

Weight Parameters	Other Parameters
S-weight		Steps
P-weight		Decay
		Intrinsic Noise
		Activation Noise
		Initial Activation
		Jolt Activation
	20%	Mixed Error Opportunities

References

The SP Model

Aphasia Modeling Project.
Dell, G. S., Lawler, E. N., Harris, H. D., & Gordon, J. K. (2004). Models of errors of omission in aphasic naming. Cognitive Neuropsychology, 21(2-4), 125-145.
Foygel, D., & Dell, G.S. (2000). Models of impaired lexical access in speech production. Journal of Memory and Language, 43, 182-216.

Bayesian SP Model

Walker, G. (2016). A Bayesian Approach to Connectionist Model Fitting. In Computational Modeling of Speech Production and Aphasia (Chapter 5, p. 97-112). Doctoral dissertation, UC Irvine: Psychology, 2016. Retrieved from: https://escholarship.org/uc/item/1cp5g22d

Contacts

Please contact Grant M. Walker for questions or comments regarding the model or the theory behind it.

Please contact Grant M. Walker for technical questions about the model's implementation, or to report a problem with this web page.

Webmaster: Grant M. Walker <grantw@uci.edu>

Last modified: Wed Aug 23 2017