HebbianConnection

HebbianConnection(
    name: str,
    source_shape: tuple,
    target_shape: tuple,
    tau: float = 10 * delta_t,
    tau_decay: float = 100 * delta_t,
    learning_rate: float = 0.1,
    learning_rule: str = "instar",
    bidirectional: bool = True,
    reward_type: str = "no_reward",
    reward_duration: tuple = (0, 1),
    wheight_reset_slot: bool = False,
)

Connects source and target activations with a learned weight tensor.

Slots

Slot	Description
Inputs	in0 source activation; in1 target activation; in2 reward signal; optional in3 reset signal
Outputs	out0 target-directed output; out1 reverse output

Import

from juniper import HebbianConnection

Notes

• Permanent buffer wheights stores learned weights.
• Buffers reward_timer and reward_onset track reward timing.