alpha_vss

Note: Only applicable in the Rescorla-Wagner mechanism.

Specifies the learning rate for vss-values (the associative strength between pairs of stimulus elements), used in the updating of vss-values.

The parameter alpha_vss can be set to certain values for certain particular element-element pairs. It can also be specified as a single value, in which case the same learning rate is used for all element-element pairs. Wildcards (*) may be used to set several values at once.

Syntax

alpha_vss = e1->e2: v1, e3->e4: v2, ..., default: d
alpha_vss = v1
alpha_vss = e1->*: v1
alpha_vss = *->e1: v1
alpha_vss = *->*: v1

where v1,v2,...,vn and d are scalar expressions.

Description

alpha_vss = e1->e2: v1, e3->e4: v2, ..., default: d sets the learning rate for

  • vss(e1->e2) to v1,

  • vss(e3->b4) to v2, …,

and the learning rate for all other vss-values to d.

  • The specification is independent of the list order:

    e1->e2: v1, e3->b4: v2, default: d

    is the same as

    e3->e4: v2, default: d, e1->e2: v1.

  • default need not be specified if all possible combinations element->element are present in the list. For example,

    stimulus_elements = e1, e2
alpha_vss = e1->e1:v11, e1->e2:v12, e2->e2:v22

alpha_vss = v1 sets the learning rate for all vss-values to v1.

  • alpha_vss = v1 is the same as alpha_vss = default: v1.

Dependencies

  • The property stimulus_elements must be specified before alpha_v.

  • Each stimulus element used in the specification of alpha_vss must be present in the parameter stimulus_elements.

  • Applicable only in the Rescorla-Wagner mechanism. (Use alpha_v in other mechanism.)

Examples

@variables x = 1
alpha_vss = element1->element2: x, element3->element4: x+1, default:x+2

sets the learning rate for vss(element1->element2) to 1 and for vss(element3->element4) to 2, and for the remaining possible element-element pairs to 3.

alpha_vss = element1->element2: 0.5, default:0

sets the initial value for v(element1->element2) to 0.5, and for the remaining possible element-element pairs to 3.

alpha_vss = 0.1

sets the learning rate for all v-values to 0.1.