diff --git a/src/bids_validator/context.py b/src/bids_validator/context.py
index 434018c..7b940a1 100644
--- a/src/bids_validator/context.py
+++ b/src/bids_validator/context.py
@@ -113,6 +113,7 @@ class Dataset:
     """A dataset object that loads properties on first access."""
 
     tree: FileTree
+    schema: Namespace
     ignored: list[str] = attrs.field(factory=list)
     subjects: Subjects = attrs.field(init=False)
 
@@ -129,14 +130,41 @@ def dataset_description(self) -> Namespace:
     @cached_property
     def modalities(self) -> list[str]:
         """List of modalities found in the dataset."""
-        ...
-        return []
+        result = set()
+
+        modalities = self.schema.rules.modalities
+        for datatype in self.datatypes:
+            for mod_name, mod_dtypes in modalities.items():
+                if datatype in mod_dtypes.datatypes:
+                    result.add(mod_name)
+
+        return list(result)
 
     @cached_property
     def datatypes(self) -> list[str]:
         """List of datatypes found in the dataset."""
-        ...
-        return []
+        return list(find_datatypes(self.tree, self.schema.objects.datatypes))
+
+
+def find_datatypes(
+    tree: FileTree, datatypes: Namespace, result: set[str] | None = None, max_depth: int = 2
+) -> set[str]:
+    """Recursively work through tree to find datatypes."""
+    if result is None:
+        result = set()
+
+    for child_name, child_obj in tree.children.items():
+        if not child_obj.is_dir:
+            continue
+
+        if child_name in datatypes.keys():
+            result.add(child_name)
+        elif max_depth == 0:
+            continue
+        else:
+            result = find_datatypes(child_obj, datatypes, result, max_depth=max_depth - 1)
+
+    return result
 
 
 @attrs.define
diff --git a/tests/test_context.py b/tests/test_context.py
index c58c6f7..0cc6c45 100644
--- a/tests/test_context.py
+++ b/tests/test_context.py
@@ -1,16 +1,33 @@
 from bids_validator import context
 from bids_validator.types.files import FileTree
 
+import pytest
 
-def test_load(examples):
+def test_load(examples, schema):
     tree = FileTree.read_from_filesystem(examples / 'synthetic')
-    ds = context.Dataset(tree)
+    ds = context.Dataset(tree, schema)
 
     assert ds.dataset_description.Name.startswith('Synthetic dataset')
     assert ds.subjects.participant_id == [f'sub-{i:02d}' for i in range(1, 6)]
     assert sorted(ds.subjects.sub_dirs) == [f'sub-{i:02d}' for i in range(1, 6)]
+    assert sorted(ds.datatypes) == ["anat", "beh", "func"]
+    assert sorted(ds.modalities) == ["beh", "mri"]
 
 
+@pytest.mark.parametrize(
+        "depth, expected",
+        [
+            (2, {"anat", "beh", "func"}),
+            (1, set())
+        ])
+def test_find_datatypes(examples, schema, depth, expected):
+    tree = FileTree.read_from_filesystem(examples / 'synthetic')
+    datatypes = schema.objects.datatypes
+
+    result = context.find_datatypes(tree, datatypes, max_depth=depth)
+
+    assert result == expected
+
 def test_fileparts(examples, schema):
     tree = FileTree.read_from_filesystem(examples / 'synthetic')