Recordings of stellar nerve activity were made during escape responses in living squid. Short-latency activation of the giant axons is triggered by light-flash stimulation that elicits a stereotyped startle-escape response and powerful jet. Many other types of stimuli produce a highly variable, delayed-escape response with strong jetting primarily controlled by a small axon motor pathway. In such cases, activation of the giant axons is not necessary for a vigorous escape jet. When they are utilized, the giant axons are not activated until well after the non-giant system initiates the escape response, and excitation is critically timed to boost the rise in intramantle pressure. Squid thus show at least two escape modes in which the giant axons can contribute in different ways to the control of a highly flexible behavior.Analyses of electrical activity in giant axons and cell bodies have contributed greatly to our understanding of neural control of behavior (1). Where giant axons exist, they most often control startle-and rapid-escape responses and are primarily responsible for the characteristics of short latency and synchronous muscle activation (2, 3). Escape swimming has been analyzed in detail in crayfish (4) and in teleosts (5), where a single action potential in a giant axon elicits a complex, but stereotyped, motor output at very short latency (<40 ms). These giant axons are not motor fibers themselves but act as interneurons to orchestrate coordinated discharge of many motoneurons.Crayfish and teleosts possess parallel, small axon pathways that can effect rapid escape responses in the absence of giant axon activity, but with a longer latency (50-500 ms; ref. 4). Interplay between giant and non-giant motor pathways during escape behavior has not been extensively studied, and such interaction may be important in maximizing flexibility of a limited motoneuron pool. For example, in teleosts a small axon system can modify the excitability of motoneurons on which Mauthner axons synapse, and a fixed command signal in a giant axon can thereby lead to different motor responses (6).Another type of interplay may involve concerted control over the simultaneous usage ofgiant and non-giant pathways. If a giant axon could be called into play in a "thoughtful," nonobligatory way during activity independently mediated by non-giant processes, its utility would be greatly enhanced.Efforts to demonstrate such a dual usage of giant axons have been equivocal, however (4, 7).We describe here experiments that shed new light on this idea. Although the squid giant axon has been extensively studied from numerous points of view, little is known about the axon's in vivo role in escape behavior. All-or-none reflex activation of escape jetting was inferred 50 years ago from studies of stellar nerve-mantle preparations (8, 9), and this idea remains widely accepted. Recordings of giant axon activity in relation to behavior have not been reported, however, and studies of escape responses have focused on biomechanical aspects (10-1...