Fig. 3: Temporal-tactile plasticity based on excitatory and inhibitory neuromorphics.

a Design of the MIPM. The schematic shows the processing pathway of tactile signals. b The postsynaptic current (PSC) curves corresponding to 10 kPa, 20 kPa, 50 kPa loading pressure. As the loading pressure increased, the absolute value of the change in postsynaptic current amplitude (ΔI) enhanced. c, d The decay curves of MIPMs’ PSCs in stiff-state and soft-state, demonstrating transition from short-term plasticity (STP) to long-term plasticity (LTP) of MIPM. e The absolute value of the change in PSC amplitude as a function of pressures applied to MIPMs in stiff and soft state. Data points represent average values from ten measurements. The bipolarized plasticity facilitates high-precision recognition of complex tactile signals. f Temporal-tactile plasticity of MIPM with two heterogel mechanogates at different stiffness states demonstrating BCM learning rule and Hebbian learning rule. BCM learning: Three controlled trials (different pressures of 10 kPa, 20 kPa and 50 kPa, respectively) as the experienced historical activities activated three non-volatile current levels from the resting current. Subsequent testing pressure (10 kPa) input yields the spike trains with the same frequency, but achieving different synaptic responses (ΔI and Δθ, where Δθ represents the time interval between the neuromorphic signal generation and its reaching to the resting current level). This result illustrates MIPM’s BCM learning function. Hebbian learning: MIPM modulated the postsynaptic current variation of subsequent pressure stimulations by controlling excitatory and inhibitory non-volatile current levels (ΔM₁ and ΔM₁’, where ΔM represents the absolute value of the change in non-volatile current) based on different historical pressures, demonstrating Hebbian learning rules.