In formal language theory, in particular in algorithmic learning theory, a class C of languages has finite thickness if every string is contained in at most finitely many languages in C. This condition was introduced by Dana Angluin as a sufficient condition for C being identifiable in the limit. ^[1]

Given a language L and an indexed class C = { L₁, L₂, L₃, ... } of languages, a member language L_j ∈ C is called a minimal concept of L within C if L ⊆ L_j, but not L ⊊ L_i ⊆ L_j for any L_i ∈ C.^[2] The class C is said to satisfy the MEF-condition if every finite subset D of a member language L_i ∈ C has a minimal concept L_j ⊆ L_i. Symmetrically, C is said to satisfy the MFF-condition if every nonempty finite set D has at most finitely many minimal concepts in C. Finally, C is said to have M-finite thickness if it satisfies both the MEF- and the MFF-condition. ^[3]

Finite thickness implies M-finite thickness.^[4] However, there are classes that are of M-finite thickness but not of finite thickness (for example, any class of languages C = { L₁, L₂, L₃, ... } such that L₁ ⊆ L₂ ⊆ L₃ ⊆ ...).