Context. Understanding how protostars accrete their mass is a centralquestion of star formation. One aspect of this is trying to understandwhether the time evolution of accretion rates in deeply embedded objectsis best characterised by a smooth decline from early to late stages orby intermittent bursts of high accretion. Aims: We createsynthetic observations of deeply embedded protostars in a largenumerical simulation of a molecular cloud, which are compared directlyto real observations. The goal is to compare episodic accretion eventsin the simulation to observations and to test the methodology used foranalysing the observations. Methods: Simple freeze-out andsublimation chemistry is added to the simulation, and syntheticC18O line cubes are created for a large number of simulatedprotostars. The spatial extent of C18O is measured for thesimulated protostars and compared directly to a sample of 16 deeplyembedded protostars observed with the Submillimeter Array. If CO isdistributed over a larger area than predicted based on the protostellarluminosity, it may indicate that the luminosity has been higher in thepast and that CO is still in the process of refreezing. Results:Approximately 1% of the protostars in the simulation show extendedC18O emission, as opposed to approximately 50% in theobservations, indicating that the magnitude and frequency of episodicaccretion events in the simulation is too low relative to observations.The protostellar accretion rates in the simulation are primarilymodulated by infall from the larger scales of the molecular cloud, anddo not include any disk physics. The discrepancy between simulation andobservations is taken as support for the necessity of disks, even indeeply embedded objects, to produce episodic accretion events ofsufficient frequency and amplitude.