Compress sensory observation	${\mathbf{x}}_t^c=f_c\left({\mathbf{x}}_t\right)$
Temporally filter sensorium	${\mathbf{x}}_t^f=\left(1-{\alpha }^f\right){\mathbf{x}}_{t-1}^f+{\alpha }^f{\mathbf{x}}_t^c$
Sensory input to hippocampus	${\tilde{\mathbf{x}}}_t={\mathbf{W}}_{tile}{w}_p{f}_n\left({\mathbf{x}}_t^f\right)$
Retrieve memory	${\mathbf{p}}_t^x=attractor\left({\tilde{\mathbf{x}}}_t,{\mathbf{M}}_{t-1}\right)$
Infer entorhinal2	${\mathbf{g}}_t\sim q_{\phi }\left({\mathbf{g}}_t|{\mathbf{p}}_t^x,{\mathbf{g}}_{t-1},{\mathbf{a}}_t\right)$
Entorhinal input to hippocampus	${\tilde{g}}_t={\mathbf{W}}_{repeat}{f}_{down}\left({\mathbf{g}}_t\right)$
Infer hippocampus	${\mathbf{p}}_t\sim \mathcal{N}\left(\cdot |\mu =f_p\left({\tilde{\mathbf{g}}}_t\cdot {\tilde{\mathbf{x}}}_t\right),\sigma =f\left({\tilde{\mathbf{x}}}_t,{\tilde{\mathbf{g}}}_t\right)\right)$
Form memory	${\mathbf{M}}_t=hebbian\left({\mathbf{M}}_{t-1},{\mathbf{p}}_t\right)$
Repeat process for next observation	$\to {\mathbf{x}}_{t+1}\to {\mathbf{g}}_{t+1}\to {\mathbf{p}}_{t+1}\cdots$