---
title: 'Parametric Right 2-Adjoint: Fibered and Categorical Views'
url: https://www.emergentmind.com/topics/parametric-right-2-adjoint
type: topic
---

# Parametric Right 2-Adjoint: Fibered and Categorical Views

“Parametric right 2-adjoint” does not have a single uniform meaning across the literature represented here. The most precise categorical formulation is fibrational: for functors between fibered slices over a locally cartesian closed category, the correct parameterized notion of right adjoint is a **local fibered right adjoint**, and these are exactly the **fibered polynomial functors** [1005.4236]. Other works use different languages while exhibiting closely related structures: Hopf parametric adjoint objects across the 2-adjunction \(\operatorname{Adj}_R(\mathbf{Cat}) \leftrightarrows \operatorname{Mnd}(\mathbf{Cat})\) [1703.05854], indexed families of right adjoints arising from operadic change in equivariant algebra [1711.11246], families of adjunctions indexed by parabolic data in \(p\)-adic representation theory [1112.6340], dependent right adjoints internalized in multimode type theory [2008.08533], half 2-adjoint equivalences in Homotopy Type Theory [2008.12433], and a free \((\infty,2)\)-categorical construction that adds right adjoints to all \(1\)-morphisms [2510.05371].

## 1. Local fibered right adjoints as the precise fibrational notion

In the formulation of Kock–Kock, the ambient structure is a locally cartesian closed category \(\mathcal E\) with terminal object, together with the **fibered slices**
\[
\mathcal E\!\mid I \to \mathcal E,\qquad \mathcal E\!\mid J \to \mathcal E,
\]
whose fiber over \(K\in \mathcal E\) is canonically
\[
(\mathcal E\!\mid I)_K \cong \mathcal E/(I\times K),\qquad (\mathcal E\!\mid J)_K \cong \mathcal E/(J\times K).
\]
This packages all slices \(\mathcal E/(I\times K)\) uniformly in the parameter \(K\), and that uniformity is the source of the “parametric” interpretation [1005.4236].

A fibered functor
\[
R:\mathcal E\!\mid I \to \mathcal E\!\mid J
\]
is called a **local fibered right adjoint** if, after the canonical factorization
\[
\mathcal F \xrightarrow{\bar R} \mathcal G/R(1_\mathcal F) \xrightarrow{\mathrm{dom}} \mathcal G,
\]
the induced \(\bar R\) is a **fibered right adjoint**. The paper stresses the difference between an ordinary local right adjoint and a local fibered right adjoint: the latter is a right adjoint in the 2-category of fibrations over \(\mathcal E\), so the right-adjoint behavior is required coherently with respect to base change, not merely fiberwise [1005.4236].

This yields the most explicit 2-categorical reading in the cited corpus. The relevant 2-category is
\[
\mathrm{Fib}_{\mathcal E},
\]
whose objects are categories fibered over \(\mathcal E\), whose \(1\)-cells are fibered functors, and whose \(2\)-cells are fibered natural transformations. In that setting, a local fibered right adjoint is precisely a right adjoint appearing after slicing over the image of the terminal object, but now in a parameterized 2-categorical environment [1005.4236].

## 2. Polynomial representation theorem

The central classification theorem is that local fibered right adjoints between fibered slices are exactly polynomial functors. A polynomial in \(\mathcal E\) is a diagram
\[
I \xleftarrow{s} E \xrightarrow{p} B \xrightarrow{t} J,
\]
and it determines the fibered polynomial functor
\[
\mathcal E\!\mid I \xrightarrow{s^*} \mathcal E\!\mid E \xrightarrow{p_*} \mathcal E\!\mid B \xrightarrow{t_!} \mathcal E\!\mid J.
\]
On the \(1\)-fiber, this recovers the ordinary polynomial functor
\[
\mathcal E/I \xrightarrow{\Delta_s} \mathcal E/E \xrightarrow{\Pi_p} \mathcal E/B \xrightarrow{\Sigma_t} \mathcal E/J,
\]
so
\[
P=\Sigma_t\,\Pi_p\,\Delta_s.
\]
Kock–Kock prove that if
\[
R:\mathcal E\!\mid I \to \mathcal E\!\mid J
\]
is a local fibered right adjoint, then
\[
R=t_!\,p_*\,s^*
\]
for some polynomial diagram \(I \xleftarrow{s} E \xrightarrow{p} B \xrightarrow{t} J\), and conversely every fibered polynomial functor is a local fibered right adjoint [1005.4236].

The proof factors \(R\) through the slice over \(R(1_I)\). Writing
\[
Q:=R(1_I),
\]
with \(Q\) of the form \(J \xleftarrow{t} B \to 1\), one obtains
\[
R=t_!\circ \bar R,
\]
where
\[
\bar R:\mathcal E\!\mid I \to \mathcal E\!\mid B
\]
is a fibered right adjoint. A separate classification theorem for fibered right adjoints then shows that
\[
\bar R=p_*\,s^*
\]
for a span \(I \xleftarrow{s} E \xrightarrow{p} B\), hence
\[
R=t_!\,p_*\,s^*.
\]
The converse is immediate from the factorization
\[
t_!\,p_*\,s^*=t_!\circ(p_*\,s^*),
\]
since \(p_*\,s^*\) has fibered left adjoint \(s_!\,p^*\) [1005.4236].

This theorem is the strongest precise answer in the cited literature to what should count as a parameterized or parametric right adjoint for slice-like constructions. It also explains why plain local right adjointness is insufficient outside \(\mathbf{Set}\): the missing ingredient is the fibered coherence across parameters \(K\) [1005.4236].

## 3. 2-adjunctions, Hopf operators, and parametric adjoint objects

A different 2-categorical approach appears in the study of the 2-adjunction
\[
\Phi_E \dashv \Psi_E : \operatorname{Adj}_R(\mathbf{Cat}) \leftrightarrows \operatorname{Mnd}(\mathbf{Cat}),
\]
where objects of \(\operatorname{Adj}_R(\mathbf{Cat})\) are adjunctions and objects of \(\operatorname{Mnd}(\mathbf{Cat})\) are monads. In this setting, the paper does not define a literal “parametric right 2-adjoint,” because the naive right-parametric object is obstructed by variance and opposition issues. Instead it introduces **Hopf parametric adjoint objects** [1703.05854].

On the adjunction side, an adjoint object exists for a \(1\)-cell
\[
(J,V,\lambda^{JV})
\]
precisely when the structural natural transformation \(\lambda^{JV}\) is invertible. On the monad side, an adjoint object for
\[
(J,\psi^J)
\]
exists precisely when \(\psi^J\) is invertible. The paper proves that these characterizations correspond across the 2-adjunction, so adjoint objects are transported between \(\operatorname{Adj}_R(\mathbf{Cat})\) and \(\operatorname{Mnd}(\mathbf{Cat})\) by invertibility of the relevant structural \(2\)-cell [1703.05854].

The parametric extension begins with an ordinary parametric adjunction
\[
F \dashv_P G : \mathcal C \to \mathcal D
\]
in the classical sense, indexed by a category \(P\). Because the naive right-parametric object does not remain a \(1\)-cell of the same 2-category, the paper imposes a **left Hopf** condition: a certain Hopf operator must be invertible. Under this condition, the left part of the parametric data induces, for each parameter, an adjoint object, and the right-side structure assembles dinaturally. The result is denoted a **Hopf parametric adjoint object** rather than a parametric right 2-adjoint [1703.05854].

The principal comparison theorem states that the 2-adjunction \(\Phi_E \dashv \Psi_E\) induces a bijection between Hopf parametric adjoint objects on the adjunction side and Hopf parametric adjoint objects on the monad side. A further lifting theorem identifies Hopf parametric adjunctions with parametric adjoint liftings to Eilenberg–Moore categories. This makes the paper one of the most direct 2-categorical sources for the topic, even though its terminology is deliberately more specialized [1703.05854].

## 4. Indexed families of right adjoints in concrete algebra and geometry

Several cited works do not formulate a theory of parametric right 2-adjoints, but they exhibit families of adjunctions indexed by an external parameter and are therefore structurally suggestive.

In equivariant algebra, for \(N_\infty\) operads \(O\subset O'\), there is a forgetful functor
\[
i_O^{O'}: O'\text{-}\mathsf{Tamb} \to O\text{-}\mathsf{Tamb}
\]
that forgets norm maps present in \(O'\) but absent from \(O\). This functor has both a left adjoint and a right adjoint, and the right adjoint is characterized objectwise by
\[
F_O(O',R)(T)\cong \Big(i_O^{O'}(A^{O'}[x_T]),R\Big),
\]
naturally in the finite \(G\)-set \(T\). The construction depends on the inclusion of indexing systems \(O\subset O'\), so there is a family of adjunctions varying over the poset of indexing systems. The paper explicitly states that it does not formulate a 2-category of indexing systems or a right 2-adjoint theorem, but it also notes that one may read the assignment \(O\mapsto O\text{-}\mathsf{Tamb}\) as suggestive of an indexed-category or pseudofunctor interpretation [1711.11246].

In the representation theory of reductive \(p\)-adic groups, for each parabolic \(P_I=L_IU_I\) with opposite \(P_I^-\), one has normalized parabolic induction \(i_I\), normalized Jacquet functors \(r_I\) and \(r_I^-\), and Bernstein’s second adjointness
\[
i_I \dashv r_I^-.
\]
The geometric construction uses a \(G\times G\)-equivariant bimodule map
\[
B_I:\mathcal S(X_I)\to \mathcal S(G)
\]
that induces a natural transformation
\[
\beta_I:i_I r_I^- \to \mathrm{Id},
\]
while the canonical map
\[
\mathrm{Can}:\mathrm{Id}\to r_I^- i_I
\]
provides the complementary adjunction datum. The paper emphasizes that this is not packaged as a theorem on parametric right 2-adjoints, but the construction is indexed by parabolic or Levi data \(I\subset E\), so it gives a family of adjunctions that is naturally interpreted bicategorically in terms of objects, \(1\)-morphisms, and \(2\)-morphisms [1112.6340].

A thin 1-categorical analogue occurs for Pultr functors on digraphs. For a Pultr template \(\mathcal T\), one always has
\[
\Lambda_{\mathcal T} \dashv \Gamma_{\mathcal T},
\]
and in many cases \(\Gamma_{\mathcal T}\) itself admits a further right adjoint
\[
\Gamma_{\mathcal T} \dashv \Omega_{\mathcal T},
\]
yielding chains
\[
\Lambda_{\mathcal T}\dashv \Gamma_{\mathcal T}\dashv \Omega_{\mathcal T}.
\]
The paper does not use 2-categorical language, and it works largely in the thin category of digraphs, but it provides a substantial body of explicit iterated right-adjoint phenomena that function as concrete lower-dimensional analogues [1304.2204].

## 5. Internal and homotopical reformulations

In type theory, the most closely related construction is the **transpension** type former. The paper introduces
\[
\transplong{u : \IU} \codot A
\]
and states that, in cartesian settings, it is right adjoint to quantification over a shape variable,
\[
\Pi(i : \IX),
\]
while in the general multimodal system the relevant adjoint chains are
\[
\sumshrt{u} \dashv \freshshrt{u} \dashv \lollishrt{u} \dashv \transpshrt{u}
\]
and, in cartesian settings,
\[
\pairshrt{u} \dashv \wknshrt{u} \dashv \funcshrt{u} \dashv \transpshrt{u}.
\]
Semantically, this is expressed by the theorem
\[
\sumpsh{\yoneda U}{\Xi} \dashv \freshpsh{\yoneda U}{\Xi} \dashv \lollipsh{\yoneda U}{\Xi} \dashv \transppsh{\yoneda U}{\Xi}.
\]
The ambient multimode type theory is parametrized by a strict 2-category of modes, modalities, and \(2\)-cells, but the paper is explicit that the transpension itself is an ordinary right adjoint functor or dependent right adjoint modality, not a higher-categorical right 2-adjoint in the usual bicategorical sense [2008.08533].

Homotopy Type Theory supplies a different use of “2-adjoint.” For a function \(f:A\to B\), the paper defines the right-flavored half 2-adjoint equivalence structure
\[
\mathsf{ish2adj}\, f
\]
and the left-flavored variant
\[
\mathsf{ish2adjl}\, f.
\]
The defining higher coherence for \(\mathsf{ish2adj}\,f\) is
\[
\mathsf{Coh}\,\eta \cdot g\llbracket \tau \rrbracket \sim \theta_f,
\]
while the left variant uses
\[
\tau_g \cdot \mathsf{Coh}\,\varepsilon \sim f\llbracket \theta \rrbracket.
\]
Both are propositions, and the paper proves
\[
\mathsf{ish2adj}\,f \leftrightarrow \mathsf{ish2adjl}\,f.
\]
By contrast, the full \(2\)-adjoint equivalence structure
\[
\mathsf{2adj}\,f
\]
is generally not a proposition. This is not a theory of parametric right 2-adjoints, but it gives a precise internal account of right- and left-flavored 2-dimensional adjunction coherence in HoTT [2008.12433].

These type-theoretic developments show that the expression “right 2-adjoint” can refer either to an external 2-categorical universal property or to an internal coherence structure on equivalences. The cited papers make that distinction explicit by reserving ordinary adjunction language for the semantics, while using “2-adjoint equivalence” for higher path coherence [2008.08533; 2008.12433].

## 6. Free addition of right adjoints and higher-categorical scope

A recent \((\infty,2)\)-categorical construction makes the adjoint-adding aspect entirely explicit. There is a functor
\[
Z_{+}^2 : Cat_{(\infty,1)} \to Cat_{(\infty,2)}
\]
such that, for an \((\infty,1)\)-category \(C\), the objects of \(Z_{+}^2(C)\) are the objects of \(C\), the \(1\)-morphisms are formal zigzags of \(1\)-morphisms of \(C\), and the \(2\)-morphisms are generated under vertical composition by formal zigzags of commutative squares, with the first and last vertical legs invertible. For every \(1\)-morphism \(f\) of \(C\), its image in \(Z_{+}^2(C)\) has a chosen right adjoint, and the paper proves the universal property
\[
Map(Z_{+}^2(C), D) \simeq Map(C, \tau_1 D^{\mathrm{ladj}}).
\]
Equivalently,
\[
Z_{+}^2 \dashv \tau_1(-)^{\mathrm{ladj}}.
\]
The example
\[
Z_{+}^2([1]) \simeq Adj
\]
identifies the zigzag construction on the walking arrow with the walking adjunction [2510.05371].

This is not presented under the label “parametric right 2-adjoint,” but it has the expected universal mapping-space behavior of a higher-categorical adjunction. A plausible implication is that it gives a free completion under right adjoints in a form that is close in spirit to a 2-categorical parametric construction, even though the paper works with complete globular double Segal spaces rather than strict 2-categories [2510.05371].

Taken together, the cited works support a layered understanding of the subject. The most exact formulation of a parameterized right adjoint is the fibrational notion of **local fibered right adjoint**, classified by polynomial functors [1005.4236]. The most explicit 2-categorical transport mechanism is given by **Hopf parametric adjoint objects** under the Adj–Mnd 2-adjunction [1703.05854]. Several major examples then realize indexed families of adjunctions without formalizing a single global right 2-adjoint theorem, while type theory and \((\infty,2)\)-category theory show how right-adjoint data can be internalized or freely generated [1711.11246; 1112.6340; 2008.08533; 2008.12433; 2510.05371]. In that sense, “parametric right 2-adjoint” names a cluster of related ideas rather than a universally fixed definition.

Source: https://www.emergentmind.com/topics/parametric-right-2-adjoint