Population Models for Star Formation Timescales in Early Galaxies: The First Step Towards Solving Outshining in Star Formation History Inference

Abstract

JWST have revealed temporarily-quenched and ultraviolet-luminous galaxies in the early universe, suggesting enhanced star formation stochasticity. Verifying this hypothesis is critical, yet challenging; outshining, wherein light from young stars dominates the spectral energy distribution, represents perhaps the greatest challenge in inferring the formation histories of unresolved galaxies. In this paper, we take a simple model of burstiness and show that state-of-the-art inference methods with flexible star formation histories (SFHs) and neutral priors, while recovering average star formation rates (SFRs; $\sim0.1$ dex median offset), fail to recover the complexities of fluctuations on tens of Myr timescales, and typically underestimate masses in bursty systems ($\sim0.15$ dex). Surprisingly, detailed SFH recovery is still sensitive to priors even when data quality is optimal, e.g., including high signal-to-noise ($\rm20~pixel^{-1}$) spectroscopy with wide coverage (rest-frame $0.12-1.06~μ$m). Crucially, however, refitting the same data with a prior correctly encoding the bursty expectation eliminates these biases: median offsets in mass and SFRs decrease to $\sim 0.04$ dex and $\sim 0.05$ dex, respectively. Under the assumption that current population burstiness predicts past SFH, the solution to outshining in modeling statistical samples is empirically measuring recent galaxy SFHs with population modeling. A prototype is H$α$/UV: while helpful, it is insufficient to constrain the expected complex burstiness. To this end, we introduce a more complete, quantitative population-level approach and demonstrate that it promises to recover the typical amplitude, timescale, and slope of the recent SFH to high accuracy. This approach thus has the strong potential to solve outshining using observations from JWST.