A neuronal network switch for approach/avoidance toggled by appetitive state

Abstract

Concrete examples of computation and implementation of cost/benefit decisions at the level of neuronal circuits are largely lacking. Such decisions are based on appetitive state, which is the integration of sensation, internal state, and memory. Value-based decisions are accessible in neuronal circuitry of simple systems. In one such system, the predatory sea slug Pleurobranchaea, appetite is readily quantified in behavior and related to approach/avoidance decision. Moreover, motor aspects of feeding and turning can be observed as fictive motor output in the isolated central nervous system (CNS). Here we found that the excitation state of the feeding motor network both manifested appetitive state and controlled expression of orienting versus avoidance. In isolated CNSs, spontaneous feeding network activity varied proportionally to donor feeding thresholds. CNSs from low- and high-feeding-threshold donors expressed fictive orienting or avoidance, respectively, in response to brief stimulation of sensory nerves. Artificially exciting the feeding network converted fictive avoidance to orienting. Thus, the feeding network embodied appetitive state and toggled approach/avoidance decision by configuring response symmetry of the premotor turn network. A resulting model suggests a basic cost/benefit decision module from which to consider evolutionary elaboration of the circuitry to serve more intricate valuation processes in complex animals.

Figure 1

Conservation of donor appetitive state in the isolated CNS. A: The isolated CNS, showing nerves stimulated and recorded in these experiments. Excitation state of the feeding network was recorded from feeding motor nerves R3. In some experiments the feeding network was excited by stimulating the stomatogastric Nerve (SGN) innervating the gut and buccal cavity, or by driving a PC_p feeding command neuron [6]. Unilateral stimulation of large oral veil nerves (LOVNs) triggered the fictive turn response recorded from ipsi- and contralateral Lateral Body Wall Nerves (LBWNs; [4]). B: Spontaneous burst frequency recorded from nerve R3 of isolated CNS’s was less from high-threshold donors than for low threshold animals. C: R3 burst frequency was an approximately linear function of donor feeding thresholds on a log-log plot (N=25; R²=0.54 and 0.59 for proboscis extension and biting, respectively). Line fits were by Least Squares. Three high threshold donor CNS’s did not show burst patterns in R3 and were excluded here.

Figure 2

Fictive turn direction was a function of donor feeding thresholds. A: CNS’s of high threshold donors responded to sensory LOVN stimulation with fictive avoidance turns (N=17) or null responses (N=12), while orienting turns (N=11) characterized CNS’s from animals with lower feeding thresholds (p<0.001 for both biting and proboscis extension thresholds, two-tailed Mann-Whitney U Test). Transition from avoidance/null to orienting responses occurred at proboscis extension thresholds of 10⁻³ and between 10⁻¹ and 10⁰ for biting. B: Fictive turn direction was represented in differing relative spike rates of the LBWNs [4] following LOVN stimulation (bar). In the two representative experiments shown, an avoidance turn (left) was seen in higher spike rates in LBWN contralateral to the stimulated nerve, while an orienting turn (right) was shown in higher relative activity in the ipsilateral LBWN (p<0.0001 in both cases, two-tailed Mann-Whitney U Tests). The initial post-stimulation peaks corresponded to fictive withdrawal preceding the fictive turn response (solid arrow) [4], as indicated.

Figure 3

Increasing feeding network excitation state switched the fictive turn from avoidance to orienting. In CNS's from high feeding threshold donors avoidance turns (A) were converted to orienting (B) when a PC_P feeding command neuron was penetrated with a microelectrode and driven to induce rhythmic bursting in the feeding motor nerve R3 (N=4). Increasing feeding network excitation by stomatogastric nerve stimulation (4 hz, 2 msec duration pulses) to drive slow rhythmic bursting in the buccal nerve R3 switched the avoidance turn (C) to orienting (D; N=5). Fictive avoidance (A, C) corresponded to higher spike frequency in contralateral LBWN (cLBWN) (p<0.0001 in both cases, Mann-Whitney U Test). Fictive orienting (B, D) matched higher activity in ipsilateral LBWN (iLBWN) (B, p<0.005; D, p<0.0001). Such significant differences were observed in all experiments. See also Figure S1.

Figure 4

A summary model for cost-benefit decision in a simple forager, summarizing appetitive regulation of approach-avoidance. A. The excitation state of the homeostatic (feeding) network controls expression of orienting vs. avoidance via corollary outputs to the directional turn motor network. The transition is sub-threshold to active feeding. B. Sensory inputs for resource quality, sensory signatures and nociception access integrating sensory networks for Incentive and Deterrence, which each promotes excitation of feeding and avoidance, respectively. Excitation in the homeostatic feeding network suppresses stimulus avoidance behavior and promotes transition to orienting approach from avoidance. Active avoidance and satiation are inhibitory to appetitive state in the homeostatic network, while homeostatic network activity also suppresses Deterrence input [cf. ref. 21]. Except where noted with minus signs, all arrows represent excitatory effect.