Two postsynaptic activity (PA) thresholds θ
min and θ
max are relevant variables for understanding plasticity in biological synapses. For PA>θ
max the synaptic weight increases. It decreases for θ
min<PA<θ
max and for PA<θ
min no significant change in synaptic weight is detected. Synaptic metaplasticity consists in the shift of these thresholds, specially θ
max, according to the actual value of the synaptic weight. If the actual weight is big, θ
max takes higher values making the window θ
min<PA<θ
max wider, and therefore priming synaptic depression over the range of postsynaptic activities. In the opposite case, when θ
max takes smaller values potentiation is favoured. We propose that a synaptic weight model, based on a conditional probability, accomplishes above properties, supporting the hypothesis of statistical computation in biological synapses
