Pings the body electric (and means it too). Focus on "interval coding. I. Burst interspike intervals as indicators of stimulus intensity" and "interval coding. II. Dendrite-dependent mechanisms".


Structure/function questions are at the forefront of research in biological sensory systems. These questions are very elegantly addressed in a pair of articles on signal representation in electric fish (Oswald et al., 2007; Doiron et al., 2007). The papers elucidate the meaning of doublets in ELL cells, and provide a mechanistic model that captures most of the functional relation. What more can one want? Two pillars in biology are the questions of function and structure – what is a particular organ doing, and how is that function implemented by the underlying biological mechanisms? These topics translate directly into two of the main questions in neural coding research – what information is represented in the brain, and how is this representation/computation implemented by the neural substrate. In a pair of companion articles, Oswald, Doiron and Maler address these questions very elegantly in an electric fish preparation. In Oswald et al. (2007), they demonstrate the function of bursts for signal representation in ELL pyramidal cells. The authors show unequivocally that a simple feature of the stimulus – the intensity of stimulus upstrokes – is associated with burst inter-spike intervals (ISI). They also demonstrate quantitatively that upstroke features (amplitude and slope) represented by a given burst ISI are highly discriminable from upstrokes associated with different burst ISIs, distinct ISI groups being about 2ms apart. Thus, bursts act as interval coders of stimulus intensity. This finding would on its own be sufficient to single out the manuscript as a major step toward clarifying functional properties of sensory neurons. But instead of stopping here, the indomitable trio forged ahead to raise the bar higher with their second manuscript (Doiron et al., 2007). In it, they proposed and tested a mechanistic model by which the neural function elucidated in the first paper can be implemented in the model system. They succeeded in correlating a specific mechanism in ELL neurons, a dendritedependent depolarizing after-potential (DAP) with the generation of the distinct bursts studied in the first manuscript. And what is more, the authors discuss the relation of all this function and mechanism to potential natural behaviors of the fish, thus approaching neuroethological nirvana. The importance of addressing function/structure questions was already obvious to Adrian in the very early stages of neurophysiological research (Adrian, 1928). With the progress of technology and newer, better instruments, the questions have been raised again and again (Lettvin et al., 1959; de Ruyter van Steveninck & Bialek, 1988), every new attempt being more precise and more informative. In the process of investigating neural function, two styles of research developed. In one, more stress was put on quantifying the presence of information in neural response patterns, mostly through information-theoretic techniques (Rieke et al., 1997). This approach provided an existence proof that there is something more in neural activity in addition to merely sets of rates; but as with any existence proof, it left the question open of what exactly that Page 1 of 3 Articles in PresS. J Neurophysiol (January 24, 2007). doi:10.1152/jn.00019.2007 Copyright © 2007 by the American Physiological Society. neural function is. As the authors remark (Oswald et al., 2007), the only way such information may be relevant is if there exists a downstream system that can decode such pattern; otherwise its meaning, as informative as it may be, is unintelligible to the rest of the system. In the complementary direction, researchers attempted to directly demonstrate specific function for all, or some, patterns of neural activity. The electric fish has been a fruitful animal model in which such questions have been addressed (Krahe & Gabbiani, 2004). Other attempts to also glean neural function in bursts (Gabbiani et al., 1996; Reinagel et al., 1999) have preceded the current work. While very interesting technically, they tend to be closer to the ”existence” proof: Their definitions of burst were quite broad, encompassing several response patterns shown to be distinct in Oswald et al. (2007). One reason we learn more in this paper is because the authors tested their hypothesis under many different conditions, weeding and narrowing down possible explanations until only a small tight region remained. Oswald, Doiron and Maler (Oswald et al., 2007; Doiron et al., 2007) leave the reader with a simple description of neural coding by ELL neurons that nevertheless captures the functionality of the system: doublets with a specific ISI encode the amplitude of a stimulus upstroke. Then, a simple leaky integrate and fire (LIF) scheme, with an added dendrite-dependent depolarizing after-potential (DAP) mechanism, suffices to produce much of the observed system properties. So where do we go from here? An obvious improvement is to take the preparation a step further and move from current injections to investigating response characteristics to actual sensory stimuli. The LIF-DAP mechanism is also sufficiently powerful that it may offer mechanistic explanation of neural function for a variety of systems. However, the main thrust of further research will need to be toward addressing more complex cases, where a single stimulus dimension and a single neuron would not suffice. The techniques developed here can be extended, and new techniques developed for these cases. Some recent tools already attempt unbiased characteristic of distinct neural code-words, applying information-theoretic methods for dimensionality reduction to the stimulus (Pillow & Simoncelli, 2006), neural responses (Reich et al., 2001; Schnitzer & Meister, 2003), or both (Dimitrov et al., 2003), in order to extract relevant stimulus/response relations. Yet they are still far from the level of precision and completeness presented here. Unlike the habitual pronouncements of the end of physics, coming every 100 years or so, no one has yet claimed an end to brain research. Yet, sometimes research progresses in leaps and bounds, and this is one of these cases, a small spot of clarity in an otherwise blurry mural. It helps us see farther, while also inspiring further work.

Extracted Key Phrases

Cite this paper

@article{Dimitrov2007PingsTB, title={Pings the body electric (and means it too). Focus on "interval coding. I. Burst interspike intervals as indicators of stimulus intensity" and "interval coding. II. Dendrite-dependent mechanisms".}, author={Alexander G. Dimitrov}, journal={Journal of neurophysiology}, year={2007}, volume={97 4}, pages={2577-8} }