Natural Unit Maps and Monoidal Units

The big idea

A monoidal unit is neutral for a specified way of combining objects. It comes with left and right isomorphisms \(I\otimes X\cong X\) and \(X\otimes I\cong X\) for every \(X\). These isomorphisms must be natural: applying a map \(h:X\to Y\) before or after removing the unit gives the same result. Changing the combining operation can change the unit even when the category stays the same.

Worked example

For sets with Cartesian product, a singleton \(I=\{*\}\) is a unit. Define

\[ \lambda_X(,x)=x,\qquad \rho_X(x,)=x. \]

Their inverses insert the fixed entry \(\). For a function \(h:X\to Y\), both routes from \((,x)\) to \(Y\) give \(h(x)\): remove \(\) and apply \(h\), or apply \(h\) to the \(X\)-coordinate and then remove \(\). This verifies the left naturality square; the right one is analogous.

For tagged union, the empty set is the unit instead. There is no empty-side element to add, so deleting the remaining tag gives a natural bijection with \(X\).

How children may show it

A learner may write element formulas with explicit inverses, draw a commuting square, or use a before-and-after diagram that deletes a fixed unit entry. The chosen operation should always be named.

Common mix-up

“Neutral object” is incomplete without “neutral for which operation?” Another common gap is proving each component is a bijection but never checking naturality. Ask what happens to a general element under both routes involving an arbitrary map \(h\).

Try it together

For ordinary number addition and multiplication, identify the two units and write the corresponding removal equations. Then compare that familiar change of unit with the singleton and empty-set units for product and tagged union of sets.