I'm pretty sure this is an easy descriptive set theory question that I'm just blanking on.

Is it consistent with large cardinals - say, with a measurable - that every (nonempty) $\Sigma^1_4$ class of reals has a $\Delta^1_4$ member?

My recollection is that there is an easy negative answer, but I can't remember the details.


This is true in $L$ by Exercise $5\,A.4$ in Moschovakis $\text{Descriptive Set Theory}$.

In fact for all $n \geq 2$, $\Delta_n^1$ is a basis for $\Sigma_n^1$. The following is a sketch:

Using the $\Delta_2^1$ well-ordering of $({}^\omega 2)^L$, you can show that $\Sigma_n^1$ has the uniformization property.

Given $A$ which is $\Sigma_n^1$, consider $\{0\} \times A$ which is still $\Sigma_n^1$. Uniformize this, to get a $\Sigma_n^1$ singleton, $A^* = \{(0, y)\}$. Then

$$n \in y \Leftrightarrow (\exists z)(n \in z \wedge (0,z) \in A^*) \Leftrightarrow (\forall z)((0,z) \in A^* \Rightarrow n \in z)$$

Therefore, $y$ is $\Delta_n^1$ and $y \in A$.

If your question is whether it is consistent relative large cardinal that "there is a measurable cardinal and $\Delta_4^1$ is a basis for $\Sigma_4^1$", this is also true:

Moschovakis 6C.6 shows that $\Delta_{2n}^1$ determinacy implies $\Delta_{2n + 2}^1$ is a basis for $\Sigma_{2n + 2}^1$.

As above, this follows from $\Delta_{2n}^1$ determinacy implying $\Sigma_{2n + 2}^1$ uniformization.

Also Martin and Solovay in $\text{A Basis Theorem for$\dots$}$ showed that if $\mu$ is a measure on a measurable cardinal, then $L[\mu][c]$ where $c$ is generic any small forcing (say Cohen forcing), then there is a $\Pi_3^1$ (in particular $\Sigma_4^1$) set without any ordinal definable reals.

So the presence of a measurable cardinal is also not enough to imply your property.

  • $\begingroup$ I think I may have misunderstood the question. Maybe you meant: is it consistent that "there exists a measurable cardinal and $\Delta_1^1$ is a basis for $\Sigma_4^1$"? $\endgroup$ – William Oct 3 '15 at 12:08
  • $\begingroup$ @William The second part of your answer is what I was looking for. My memory, it seems, is far off (especially at midnight :P). Thanks! $\endgroup$ – Noah Schweber Oct 3 '15 at 20:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.