How do neural networks process information, store memory and generate behavior?
Support Media: Escape Swim Behavior Movie
My laboratory investigates these issues using an array of techniques, including behavioral studies, intracellular electrophysiology, optical recording of network activity with voltage sensitive dyes, and realistic computer simulations of the networks under study. Our experimental preparation is an invertebrate model system, the marine mollusk Tritonia diomedea (below, left). This animal’s large, individually identifiable neurons (below, right) make it possible to return to the same specific neurons and synapses from preparation to preparation, allowing us to dissect their roles in network function. By focusing on issues common to all animals, we seek general principles of how networks of neurons perform their functions. Issues under investigation include the neural basis of decision-making, learning and memory, network multifunctionality, and prepulse inhibition.
Our experimental preparation
Upon being touched by the tube feet of predatory seastars, Tritonia
launches its escape swim, consisting of a rhythmic series of ventral and dorsal whole-body flexions (below, left and see Escape Swim Behavior Movie). The circuit mediating this escape response (below, right) is one of the better understood in any animal, in part because the neural motor program it generates can be elicited and studied in successively reduced preparations, from the intact animal to the isolated brain (see Intact Animal Electrophysiology Movie).
|Tritonia in mid-escape from its
predator, the seastar Pycnopodia helianthoides.
At the top are simultaneous intracellular recordings from 3 central
pattern generator interneurons during a 3 cycle escape swim motor
Diagram of the Tritonia escape swim circuit.
Support Media: Intact Animal Electrophysiology Movie
Neural Basis of Decision-Making
Upon detecting a mildly aversive stimulus, Tritonia takes its time while deciding whether or not to launch an escape response. Here we study the network processes mediating this decision-making process, which, as in humans, is strongly affected by experience encoded through learning. Our studies have identified a command structure for the Tritonia's decision-making process, and a potent role for plasticity in positive-feedback pathways. Relevant publications: Katz, P.S., et al., 1994; Frost and Katz, 1996; Katz and Frost, 1997; Frost, et al., 2001.
Learning and Memory
Tritonia are very sensitive to past experience. An initial encounter with its predator induces an hour-long period of sensitization, during which a second encounter elicits swims having a lower threshold, shorter onset latency and higher number of flexion cycles. Repeated stimuli lead to behavioral habituation, the memory for which can last several days. We are studying the alterations in the swim network that encode these universal forms of nonassociative learning. Relevant publications: Frost, et al., 1996; Frost, et al., 1998; Mongeluzi and Frost, 2000; Wang and Frost, 2002; Popescu, et al., 2002; Frost and Wang, 2003; Frost, et al., 2006.
Many neural networks are multifunctional -- they participate in multiple behavioral motor programs. We are interested in the mechanisms by which single networks can manage such versatility without confusion. The Tritonia swim network, for example, participates in three behaviors, crawling, swimming, and reflexive withdrawals. Relevant publications: Popescu and Frost, 2002.
In prepulse inhibition (PPI), startle responses to sudden, unexpected stimuli are markedly attenuated if immediately preceded by a weak stimulus of almost any modality. This experimental paradigm exposes a potent inhibitory process, present in nervous systems from invertebrates to humans, that is widely considered to play an important role in reducing distraction during the processing of sensory input. The neural mechanisms mediating PPI are of considerable interest given evidence linking PPI deficits with some of the cognitive disorders of schizophrenia. In our laboratory we described the first example of PPI in an invertebrate, and have more recently identified the first specific cellular mechanisms for PPI in any organism. Relevant publications: Mongeluzi, et al., 1998; Frost, et al., 2003; Lee, et al., 2003.
Our research involves several techniques, including behavioral studies, intracellular electrophysiology, realistic network modeling (Lieb and Frost, 1997; Frost, et al., 1997; Calin-Jagemin, et al., 2006) and most recently, optical recording of network activity with voltage sensitive dyes (Frost, et al., 2007; Hill, et. al., 2010; Frost, et. al., 2010).
Optical recording of the pedal ganglion during an escape swim motor program. Each trace in the left panel displays the activity recorded from one of the 464 photodiodes of the array, which appears as the green hexagon of short line segments in the right panel, superimposed over an image of the ganglion. The positions of the optically recorded neurons are indicated with arrows.
Traditional network studies using intracellular electrodes are limited by their inability to record from more than about 4 neurons simultaneously. A powerful new tool in our laboratory to overcome this limited view of network activity is optical recording, which allows us to record from dozens to hundreds of neurons simultaneously. We use a 464 element photodiode array in combination with a voltage sensitive dye to image the firing activity of entire networks of Tritonia neurons. To make it easier to integrate conventional intracellular recording into our optical recordings, we have assembled a novel hybrid microscope, part compound, part stereo. By superimposing a photograph of the ganglion with the optical array data we can determine the exact ganglion location of the neurons being recorded optically, which then guides our placement of intracellular electrodes into the neurons of interest. We are currently using this approach to identify new members of the Tritonia swim network.