Fig. 2.

Synaptic weight updates based on to Eq. 14. Inspired by Izhikevich’s work [71]. This figure illustrates the dynamics of synaptic changes. Red lines represent spike traces from presynaptic neurons, while blue lines indicate those from postsynaptic neurons. The eligibility trace, formulated in Eq. 13, is depicted by the green line. The modulator M(t), present at times $t_1$, $t_2$, and $t_3$, modulates the synaptic weight by interacting with the eligibility trace (purple line). Both the pre- and post-traces have a temporal memory span of approximately 100 ms, akin to Fig. 1A, while the eligibility trace has a longer memory duration of approximately 5 s, allowing for delayed rewards. The eligibility trace is dynamically updated through sequential occurrences of LTP and LTD events. At $t_1$, the weight increases due to the multiplication of the eligibility trace by a positive modulatory factor. At $t_2$, a negative modulator decreases the weight, and at $t_3$, the smaller magnitude of the modulator results in only a slight modification to the synaptic weight