This page contains the results of ten replications of all simulations reported in the paper. The results are qualitatively the same as those reported, with the exception of one datapoint in one of the replications.

Run 1
Run 2
Run 3
Run 4
Run 5
Run 6
Run 7
Run 8
Run 9
Run 10

Run 1

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 2

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:
In this replication, at epoch 17 only, the interference training does eliminate the effect of earlier training on the novel word: catastrophic interference. This does not occur in the other replications. Catastrophic interference results when a previously-trained associative weight is selected probabilistically for update during subsequent training. This happens very rarely, since the distribution from which the selection is made centers on the new form and referent, which are some distance from the previously-trained ones.

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 3

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 4

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 5

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 6

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 7

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 8

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 9

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label:

Run 10

Training error:

Form attention weights:

Meaning attention weights:

Number of nodes:

One-trial learning, 10-trial learning, resistance to interference:

Comprehension and production:

Learning phonologically similar words:

Shape bias:

Learning a second label: