I have checked it but not found.  So I accept Brooks' paper is the first one covering the reactive behaviour paradigm...

It depends on what you mean. For example, with respect to theories about human movement control, the perception-action loop (we sense, we plan, we act, then repeat) has been described at least as early as 1899 by Robert Sessions Woodworth in his seminal work "The Accuracy of Voluntary Movement" (see https://archive.org/details/accuracyvolunta00woodgoog).  This work was based on an idea of continuous closed-loop control. The first consolidated theory involving feedback in both learning and control was later provided in 1971 by Adams in his paper "A closed-loop theory of motor learning" (Journal of Motor Behavior , 3 (2), 111–149). Based on engineering principles and consistent with pre-existing knowledge on the role of feedback in learning and performance (e.g., Adams, 1968), the theory proposed that movement was controlled by continuous comparison between afferent information and a set of “sensory consequences” stored in memory from previous successful movements

However, in work related to the control of goal-seeking war missiles dating back to World War II, the famous scientist Kenneth J. W. Craik showed in his seminal papers that the human operator behaves like an error-correcting servomechanism in tracking tasks. The importance of error correction, and hence of feedback, was substantiated by Craik’s colleague Vince (1948) in discrete aimed movements as well as in tracking tasks requiring continuous movement (Craik, K. J. W. (1947). Theory of the human operator in control systems: I. The operator as an engineering system. British Journal of Psychology, 38 , 56–61. Craik, K. J. W. (1948). Theory of the human operator in control systems: II. Man as an element in a control system. British Journal of Psychology, 38 , 142–148. Vince, M. A. (1948). The intermittency of control movements and the psychological refractory period. British Journal of Psychology, 38 , 149–157). 

Then, of course, one must not forget Norbert Wiener, the originator of cybernetics, in which the fundamental principle is that of feedback. His work also dates back to World War II and his work on the automatic aiming and firing of anti-aircraft guns.

The list goes on, both within the fields of cybernetics, robotics, control systems, automation and the like, as well as artificial intelligence. 

All of the work here has been highly influential in robotics and it seems totally misplaced to suggest that the "sense, plan, act" paradigm was introduced as late as 1986!

As far as I can recall the early work includes GN Saridis, Chandrasekhar, Barbara Hays-Roth et al. All these researchers explored models of human cognition from different perspectives. But pretty much talking about the same model Perception-Processing- Actuation. I also did a survey of the models in my PhD thesis.

@ARTICLE{1087032,

author={R. Brooks},

journal={IEEE Journal on Robotics and Automation},

title={A robust layered control system for a mobile robot},

year={1986},

doi={10.1109/JRA.1986.1087032},

}