diff --git a/psyneulink/core/compositions/composition.py b/psyneulink/core/compositions/composition.py
index 38627b7adb..5f4b24dfc2 100644
--- a/psyneulink/core/compositions/composition.py
+++ b/psyneulink/core/compositions/composition.py
@@ -333,12 +333,13 @@
 ^^^^^^^^^^^^
 
 `BIAS` `NodeRole` can be used to implement BIAS Nodes, which add a bias (constant value) to the input of another
-Node, that can also be modified by `learning <Composition_Learning>`. A bias Node is implemented by adding a
-`ProcessingMechanism` to the Composition and requiring it to have the `BIAS` `NodeRole`. The ProcessingMechanims
-cannot have any afferent Projections, and should project to the `InputPort` containing the values to be biased. If
-the bias is to be learned, the `learnable <MappingProjection.learnable>` attribute of the MappingProjeciton
-should be set to True. The value of the bias, and how it is applied to the values being biased are specified as
-described below:
+Node, that can also be modified by `learning <Composition_Learning>`. A BIAS Node is implemented by adding a
+`ProcessingMechanism` to the Composition and requiring it to have the `BIAS` `NodeRole`. This can be done using
+any of the methods described `above <Composition_Nodes>` for assigning `NodeRoles <NodeRole>` tp a Node. The
+ProcessingMechanism cannot have any afferent Projections, and should project to the `InputPort` of the Node with
+the values to be biased. If the bias is to be learned, the `learnable <MappingProjection.learnable>` attribute of
+the MappingProjeciton should be set to True. The value of the bias, and how it is applied to the values being biased
+are specified as described below:
 
     *Single bias value*. To apply a single scalar bias value to all elements of the array being biased, the
     `default_variable <Component_Variable>` of the BIAS Node should be specified as a scalar value, and the `matrix
diff --git a/psyneulink/library/compositions/pytorchwrappers.py b/psyneulink/library/compositions/pytorchwrappers.py
index ab8ce37e7c..9e95a04032 100644
--- a/psyneulink/library/compositions/pytorchwrappers.py
+++ b/psyneulink/library/compositions/pytorchwrappers.py
@@ -232,8 +232,10 @@ def __init__(self,
                                                        self._composition._get_node_index(node),
                                                        device,
                                                        context=context)
-                pytorch_node._is_bias = any(input_port.default_input == DEFAULT_VARIABLE
-                                            for input_port in node.input_ports)
+                # pytorch_node._is_bias = all(input_port.default_input == DEFAULT_VARIABLE
+                #                             for input_port in node.input_ports)
+                pytorch_node._is_bias = node in self._composition.get_nodes_by_role(NodeRole.BIAS)
+
             self.nodes_map[node] = pytorch_node
             self.wrapped_nodes.append(pytorch_node)
 
diff --git a/tests/composition/test_composition.py b/tests/composition/test_composition.py
index 3f898f11fe..ce0c93492f 100644
--- a/tests/composition/test_composition.py
+++ b/tests/composition/test_composition.py
@@ -8139,11 +8139,11 @@ def test_danglingControlledMech(self):
     @pytest.mark.parametrize(
         'removed_nodes, expected_dependencies',
         [
-            # (['A'], {'B': set(), 'C': set('B'), 'D': set('C'), 'E': set('C')}),
-            # (['C'], {'A': set(), 'B': set(), 'D': set(), 'E': set()}),
-            # (['E'], {'A': set(), 'B': set(), 'C': {'A', 'B'}, 'D': set('C')}),
-            # (['A', 'B'], {'C': set(), 'D': set('C'), 'E': set('C')}),
-            # (['D', 'E'], {'A': set(), 'B': set(), 'C': {'A', 'B'}}),
+            (['A'], {'B': set(), 'C': set('B'), 'D': set('C'), 'E': set('C')}),
+            (['C'], {'A': set(), 'B': set(), 'D': set(), 'E': set()}),
+            (['E'], {'A': set(), 'B': set(), 'C': {'A', 'B'}, 'D': set('C')}),
+            (['A', 'B'], {'C': set(), 'D': set('C'), 'E': set('C')}),
+            (['D', 'E'], {'A': set(), 'B': set(), 'C': {'A', 'B'}}),
             (['A', 'B', 'C', 'D', 'E'], {}),
         ]
     )