Built-in models and likelihoods

hssm.HSSM(model=...) accepts the built-in model names below. The table is tested against hssm.list_models() and each model's default configuration, so a code change that leaves this catalog stale fails CI.

When loglik_kind is omitted, HSSM prefers analytical, then approx_differentiable, then blackbox among the kinds configured for that model.

Model	Available likelihood kinds	Default	Parameters	Choices
ddm	analytical, approx_differentiable, blackbox	analytical	v, a, z, t	-1, 1
ddm_sdv	analytical, approx_differentiable, blackbox	analytical	v, a, z, t, sv	-1, 1
full_ddm	blackbox	blackbox	v, a, z, t, sz, sv, st	-1, 1
angle	approx_differentiable	approx_differentiable	v, a, z, t, theta	-1, 1
levy	approx_differentiable	approx_differentiable	v, a, z, alpha, t	-1, 1
ornstein	approx_differentiable	approx_differentiable	v, a, z, g, t	-1, 1
weibull	approx_differentiable	approx_differentiable	v, a, z, t, alpha, beta	-1, 1
race_no_bias_angle_4	approx_differentiable	approx_differentiable	v0, v1, v2, v3, a, z, t, theta	0, 1, 2, 3
ddm_seq2_no_bias	approx_differentiable	approx_differentiable	vh, vl1, vl2, a, t	0, 1, 2, 3
gamma_drift	approx_differentiable	approx_differentiable	v, a, z, t, shape, scale, c	-1, 1
lba3	analytical	analytical	A, b, v0, v1, v2	0, 1, 2
lba4	analytical	analytical	A, b, v0, v1, v2, v3	0, 1, 2, 3
lba2	analytical	analytical	A, b, v0, v1	0, 1
racing_diffusion_3	analytical	analytical	A, b, v0, v1, v2, t	0, 1, 2
poisson_race	analytical	analytical	r1, r2, k1, k2, t	-1, 1
softmax_inv_temperature_2	analytical	analytical	beta, logit1	-1, 1
softmax_inv_temperature_3	analytical	analytical	beta, logit1, logit2	0, 1, 2

Specialized model families

The table covers the hssm.HSSM(model=...) registry. Two specialized public classes have separate configuration and discovery surfaces:

• hssm.aDDM models attentional evidence accumulation from trial and fixation data.
• hssm.RLSSM combines reinforcement-learning updates with an SSM observation model; its class-level and module-level list_models functions enumerate the RL registry.

Use hssm.list_models to discover the built-in HSSM names programmatically. Custom HSSM configurations can be added with hssm.register_model.

For the conceptual difference between the likelihood kinds, see Likelihood kinds in HSSM.