minor revision to double-exp syn pointing, mods to modeling docs

Author: Alexander Ororbia

docs/modeling/neurons.md

@@ -86,6 +86,22 @@ and `dmu` is the first derivative with respect to the mean parameter.
     :noindex:
 ```
 
+#### Bernoulli Error Cell
+
+This cell is (currently) fixed to be a (factorized) multivariate Bernoulli cell. 
+Concretely, this cell implements compartments/mechanics to facilitate Bernoulli 
+log likelihood error calculations. 
+
+```{eval-rst}
+.. autoclass:: ngclearn.components.BernoulliErrorCell
+  :noindex:
+
+  .. automethod:: advance_state
+    :noindex:
+  .. automethod:: reset
+    :noindex:
+```
+
 ## Spiking Neurons
 
 These neuronal cells exhibit dynamics that involve emission of discrete action
@@ -117,10 +133,42 @@ negative pressure on the membrane potential values at `t`).
     :noindex:
 ```
 
+### The IF (Integrate-and-Fire) Cell
+
+This cell (the simple "integrator") models dynamics over the voltage `v`. Note that `thr` is used as the membrane potential threshold and no adaptive threshold mechanics are implemented for this cell model. 
+(This cell is primarily a faster, convenience formulation that omits the leak element of the LIF.)
+
+```{eval-rst}
+.. autoclass:: ngclearn.components.IFCell
+  :noindex:
+
+  .. automethod:: advance_state
+    :noindex:
+  .. automethod:: reset
+    :noindex:
+```
+
+### The Winner-Take-All (WTAS) Cell
+
+This cell models dynamics over the voltage `v` as a simple instantaneous 
+softmax function of the electrical current input, where only a single 
+spike, which wins the competition across the group of neuronal units 
+within this component, emits a pulse/spike.  
+
+```{eval-rst}
+.. autoclass:: ngclearn.components.WTASCell
+  :noindex:
+
+  .. automethod:: advance_state
+    :noindex:
+  .. automethod:: reset
+    :noindex:
+```
+
 ### The LIF (Leaky Integrate-and-Fire) Cell
 
 This cell (the "leaky integrator") models dynamics over the voltage `v`
-and threshold shift `thrTheta` (a homeostatic variable). Note that `thr`
+and threshold shift `thr_theta` (a homeostatic variable). Note that `thr`
 is used as a baseline level for the membrane potential threshold while
 `thrTheta`  is treated as a form of short-term plasticity (full
 threshold is: `thr + thrTheta(t)`).

docs/modeling/synapses.md

@@ -1,17 +1,7 @@
 # Synapses
 
-The synapse is a key building block for connecting/wiring together the various
-component cells that one would use for characterizing a biomimetic neural system.
-These particular objects are meant to perform, per simulated time step, a
-specific type of transformation -- such as a linear transform or a 
-convolution -- utilizing their underlying synaptic parameters.
-Most times, a synaptic cable will be represented by a set of matrices (or filters) 
-that are used to conduct a projection of an input signal (a value presented to a
-pre-synaptic/input compartment) resulting in an output signal (a value that
-appears within one of its post-synaptic compartments). Notably, a synapse component is
-typically associated with a local plasticity rule, e.g., a Hebbian-type
-update, that either is triggered online, i.e., at some or all simulation time
-steps, or by integrating a differential equation, e.g., via eligibility traces.
+The synapse is a key building block for connecting/wiring together the various component cells that one would use for characterizing a biomimetic neural system. These particular objects are meant to perform, per simulated time step, a specific type of transformation -- such as a linear transform or a convolution -- utilizing their underlying synaptic parameters. Most times, a synaptic cable will be represented by a set of matrices (or filters) that are used to conduct a projection of an input signal (a value presented to a pre-synaptic/input compartment) resulting in an output signal (a value that appears within one of its post-synaptic compartments). There are three general groupings of synaptic components in ngc-learn: 1) non-plastic static synapses (only perform fixed transformations of input signals); 2) non-plastic dynamic synapses (perform time-varying, input-dependent transformations on input signals); and 3) plastic synapses that carry out long-term evolution. 
+Notably, plastic synapse components are typically associated with a local plasticity rule, e.g., a Hebbian-type update, that either is triggered online, i.e., at some or all simulation time steps, or by integrating a differential equation, e.g., via eligibility traces.
 
 ## Non-Plastic Synapse Types
 
@@ -74,6 +64,20 @@ This (chemical) synapse performs a linear transform of its input signals. Note t
     :noindex:
 ```
 
+### Double-Exponential Synapse
+
+This (chemical) synapse performs a linear transform of its input signals. Note that this synapse is "dynamic" in the sense that its efficacies are a function of their pre-synaptic inputs; there is no inherent form of long-term plasticity in this base implementation. Synaptic strength values can be viewed as being filtered/smoothened through a doubleexpoential / difference of two exponentials kernel.
+
+```{eval-rst}
+.. autoclass:: ngclearn.components.DoubleExpSynapse
+  :noindex:
+
+  .. automethod:: advance_state
+    :noindex:
+  .. automethod:: reset
+    :noindex:
+```
+
 ### Alpha Synapse
 
 This (chemical) synapse performs a linear transform of its input signals. Note that this synapse is "dynamic" in the sense that its efficacies are a function of their pre-synaptic inputs; there is no inherent form of long-term plasticity in this base implementation. Synaptic strength values can be viewed as being filtered/smoothened through a kernel that models more realistic rise and fall times of synaptic conductance..

docs/ngclearn_papers.md

@@ -17,12 +17,14 @@ from data streams." arXiv preprint arXiv:1908.08655 (2019).
 a hyperdimensional predictive processing cognitive architecture."
 Proceedings of the Annual Meeting of the Cognitive Science Society (CogSci), Volume 44 (2022).
 
-5. Ororbia, A., and Kelly, M. Alex. "“Learning using a Hyperdimensional Predictive Processing Cognitive
-Architecture." 15th International Conference on Artificial General Intelligence (AGI) (2022).
+5. Ororbia, A., and Kelly, M. Alex. "Learning using a hyperdimensional predictive processing cognitive
+architecture." 15th International Conference on Artificial General Intelligence (AGI) (2022).
 
 6. Ororbia, A., Mali, A., Kifer, D., & Giles, C. L. "Lifelong neural predictive coding: Learning cumulatively online without 
 forgetting." Thirty-sixth Conference on Neural Information Processing Systems (NeurIPS) (2022).
 
+7. Ororbia, A., Friston, K., Rao, Rajesh P. N. "Meta-representational predictive coding: Biomimetic self-supervised learning." arXiv preprint arXiv:2503.21796 (2025).  
+
 <b>Note:</b> Please let us know if your work uses ngc-learn so we can update this page to accurately track 
 ngc-learn's use and include your work in the accumulating body of work in predictive processing 
 and/or brain-inspired computational modeling.

ngclearn/components/__init__.py

@@ -39,7 +39,7 @@
 from .synapses.hebbian.BCMSynapse import BCMSynapse
 from .synapses.STPDenseSynapse import STPDenseSynapse
 from .synapses.exponentialSynapse import ExponentialSynapse
-from .synapses.doubleExpSynapse import DoupleExpSynapse
+from .synapses.doubleExpSynapse import DoubleExpSynapse
 from .synapses.alphaSynapse import AlphaSynapse
 
 ## point to convolutional component types

ngclearn/components/synapses/__init__.py

@@ -1,11 +1,10 @@
 from .denseSynapse import DenseSynapse
 from .staticSynapse import StaticSynapse
 
-
 ## short-term plasticity components
 from .STPDenseSynapse import STPDenseSynapse
 from .exponentialSynapse import ExponentialSynapse
-from .doubleExpSynapse import DoupleExpSynapse
+from .doubleExpSynapse import DoubleExpSynapse
 from .alphaSynapse import AlphaSynapse
 
 ## dense synaptic components
@@ -15,7 +14,6 @@
 from .hebbian.eventSTDPSynapse import EventSTDPSynapse
 from .hebbian.BCMSynapse import BCMSynapse
 
-
 ## conv/deconv synaptic components
 from .convolution.convSynapse import ConvSynapse
 from .convolution.staticConvSynapse import StaticConvSynapse
@@ -26,7 +24,6 @@
 from .convolution.hebbianDeconvSynapse import HebbianDeconvSynapse
 from .convolution.traceSTDPDeconvSynapse import TraceSTDPDeconvSynapse
 
-
 ## modulated synaptic components
 from .modulated.MSTDPETSynapse import MSTDPETSynapse
 # from .modulated.REINFORCESynapse import REINFORCESynapse

ngclearn/components/synapses/doubleExpSynapse.py

@@ -6,7 +6,7 @@
 from ngcsimlib.compartment import Compartment
 from ngcsimlib.parser import compilable
 
-class DoupleExpSynapse(DenseSynapse): ## dynamic double-exponential synapse cable
+class DoubleExpSynapse(DenseSynapse): ## dynamic double-exponential synapse cable
     """
     A dynamic double-exponential synaptic cable; this synapse evolves according to difference of two exponentials
     synaptic conductance dynamics.