The perils of omitting omissions when modeling evidence accumulation

Response deadlines are commonly imposed in decision-making research to incentivize speedy decisions and sustained attention. This procedure frequently leads to a proportion of trials in which no response is made, and these omissions are often simply removed from the data during analysis. Here we show that this seemingly trivial assumption is in fact quite consequential for parameter estimation. We propose that omissions should instead be treated as observations to inform inference of the underlying generative process that led to their occurrence. Using new tools from likelihood-free inference applicable to a broad class of sequential sampling models (SSMs), we enable fast computation of omission probability without explicit integration, and clarify the degree to which omitting omissions – even in seemingly benign settings – can lead researchers astray. We explore this phenomenon in the setting of SSMs with constant and time-varying boundaries, and show that parameter recovery is improved by incorporating a model of omission probability. We show that the benefits of modeling omission probability are distinct from benefits gained from past approaches of modeling attentional lapses to account for these omissions. Our findings shine a light on the consequences of omitting omission in modeling choice behavior with SSMs, and demonstrate how joint modeling of observed and omitted data can improve parameter inference and therefore the reliability of downstream scientific conclusions.Author summary: Many studies of human decision-making use tasks that impose time limits, but researchers often ignore trials in which participants fail to respond before the deadline. This paper shows that such “omissions” are more informative than they might appear—and that leaving them out can lead to misleading conclusions. The authors demonstrate that ignoring omissions distorts the estimated parameters that describe how people accumulate evidence and make choices under time pressure. To address this, they develop a new computational approach that models both responses and omissions together. Using modern machine-learning tools, specifically Likelihood Approximation Networks (LANs) combined with a new Omission Probability Network (OPN), the method efficiently estimates how likely omissions are to occur under different model settings. Across a range of decision-making models and conditions, this joint modeling approach greatly improves the accuracy and reliability of parameter estimation, even when omissions are rare. The work highlights how small analytic shortcuts—like excluding no-response trials—can meaningfully affect scientific inference and proposes a practical solution implemented in an open-source software package, making it easier for researchers to include omissions in their models and produce more trustworthy conclusions about the mechanisms of human decision-making.