Limits...
On the precarious path of reverse neuro-engineering.

Marom S, Meir R, Braun E, Gal A, Kermany E, Eytan D - Front Comput Neurosci (2009)

Bottom Line: We demonstrate that application of reverse engineering to the study of the design principle of a functional neuro-system with a known mechanism, may result in a perfectly valid but wrong induction of the system's design principle.If in the very simple setup we bring here (static environment, primitive task and practically unlimited access to every piece of relevant information), it is difficult to induce a design principle, what are our chances of exposing biological design principles when more realistic conditions are examined?Implications to the way we do Biology are discussed.

View Article: PubMed Central - PubMed

Affiliation: Network Biology Research Laboratories, Technion - Israel Institute of Technology Haifa, Israel. marom@technion.ac.il

ABSTRACT
In this perspective we provide an example for the limits of reverse engineering in neuroscience. We demonstrate that application of reverse engineering to the study of the design principle of a functional neuro-system with a known mechanism, may result in a perfectly valid but wrong induction of the system's design principle. If in the very simple setup we bring here (static environment, primitive task and practically unlimited access to every piece of relevant information), it is difficult to induce a design principle, what are our chances of exposing biological design principles when more realistic conditions are examined? Implications to the way we do Biology are discussed.

No MeSH data available.


Trajectory of the agent's path, over 1500 s, in an obstacle avoidance task. Obstacles and walls are depicted in gray. Inputs from the two ultrasonic eyes of a Lego Mindstorms vehicle are sampled at 0.2 Hz and translated into stimulation of a large random network of cortical neurons at two different sites. The side corresponding to the nearest visual object (relative to the vehicle's longitudinal axis) is classified using an Edit-distance metric based on the recruitment order of 8 neurons, similar to procedures shown in Figure 6 of Shahaf et al. (2008). Based on the classified activity, a command is sent to the appropriate motor attached to one of the wheels. See Video S1 in Shahaf et al. (2008) for technical details.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2691154&req=5

Figure 1: Trajectory of the agent's path, over 1500 s, in an obstacle avoidance task. Obstacles and walls are depicted in gray. Inputs from the two ultrasonic eyes of a Lego Mindstorms vehicle are sampled at 0.2 Hz and translated into stimulation of a large random network of cortical neurons at two different sites. The side corresponding to the nearest visual object (relative to the vehicle's longitudinal axis) is classified using an Edit-distance metric based on the recruitment order of 8 neurons, similar to procedures shown in Figure 6 of Shahaf et al. (2008). Based on the classified activity, a command is sent to the appropriate motor attached to one of the wheels. See Video S1 in Shahaf et al. (2008) for technical details.

Mentions: We use a biological toy model, a realized Braitenberg Vehicle II (Braitenberg, 1984). This is a continuously moving Lego robot that is equipped with two ultrasonic eyes that transmit their input to a large scale network of real, cultured biological cortical neurons (for review see Marom and Shahaf, 2002). The task of the agent (the Lego apparatus together with the biological network) is to avoid running into obstacles in a static environment. Based on the electrical responses of neurons to the input from the ultrasonic eyes, a decision is taken (by a well-defined algorithm) as to which direction should the agent be driven (see caption of Figure 1). This algorithm considers only the delay from stimulus time to first spike that is emitted by broadly-tuned neurons (i.e. neurons that responded to input from the right as well as the left ultrasonic eyes). The responding neurons are ranked based on the time to first spike, and the resulting rank order represents the input source. The algorithm, which is based on a reported analysis of response dynamics (for detailed explanation see Shahaf et al., 2008), performs flawlessly in spatial input classification tasks. This is demonstrated in a movie file (Supportive Information Video S1 in Shahaf et al., 2008) that shows the behaviour of the agent over 1500 s; Figure 1 depicts the trajectory of the system over that period of time. The agent performs perfectly in the sense that it succeeds in its avoidance task. Importantly, no learning is involved; the representations of stimuli from the ultrasonic eyes are fixed by the rank-order.


On the precarious path of reverse neuro-engineering.

Marom S, Meir R, Braun E, Gal A, Kermany E, Eytan D - Front Comput Neurosci (2009)

Trajectory of the agent's path, over 1500 s, in an obstacle avoidance task. Obstacles and walls are depicted in gray. Inputs from the two ultrasonic eyes of a Lego Mindstorms vehicle are sampled at 0.2 Hz and translated into stimulation of a large random network of cortical neurons at two different sites. The side corresponding to the nearest visual object (relative to the vehicle's longitudinal axis) is classified using an Edit-distance metric based on the recruitment order of 8 neurons, similar to procedures shown in Figure 6 of Shahaf et al. (2008). Based on the classified activity, a command is sent to the appropriate motor attached to one of the wheels. See Video S1 in Shahaf et al. (2008) for technical details.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2691154&req=5

Figure 1: Trajectory of the agent's path, over 1500 s, in an obstacle avoidance task. Obstacles and walls are depicted in gray. Inputs from the two ultrasonic eyes of a Lego Mindstorms vehicle are sampled at 0.2 Hz and translated into stimulation of a large random network of cortical neurons at two different sites. The side corresponding to the nearest visual object (relative to the vehicle's longitudinal axis) is classified using an Edit-distance metric based on the recruitment order of 8 neurons, similar to procedures shown in Figure 6 of Shahaf et al. (2008). Based on the classified activity, a command is sent to the appropriate motor attached to one of the wheels. See Video S1 in Shahaf et al. (2008) for technical details.
Mentions: We use a biological toy model, a realized Braitenberg Vehicle II (Braitenberg, 1984). This is a continuously moving Lego robot that is equipped with two ultrasonic eyes that transmit their input to a large scale network of real, cultured biological cortical neurons (for review see Marom and Shahaf, 2002). The task of the agent (the Lego apparatus together with the biological network) is to avoid running into obstacles in a static environment. Based on the electrical responses of neurons to the input from the ultrasonic eyes, a decision is taken (by a well-defined algorithm) as to which direction should the agent be driven (see caption of Figure 1). This algorithm considers only the delay from stimulus time to first spike that is emitted by broadly-tuned neurons (i.e. neurons that responded to input from the right as well as the left ultrasonic eyes). The responding neurons are ranked based on the time to first spike, and the resulting rank order represents the input source. The algorithm, which is based on a reported analysis of response dynamics (for detailed explanation see Shahaf et al., 2008), performs flawlessly in spatial input classification tasks. This is demonstrated in a movie file (Supportive Information Video S1 in Shahaf et al., 2008) that shows the behaviour of the agent over 1500 s; Figure 1 depicts the trajectory of the system over that period of time. The agent performs perfectly in the sense that it succeeds in its avoidance task. Importantly, no learning is involved; the representations of stimuli from the ultrasonic eyes are fixed by the rank-order.

Bottom Line: We demonstrate that application of reverse engineering to the study of the design principle of a functional neuro-system with a known mechanism, may result in a perfectly valid but wrong induction of the system's design principle.If in the very simple setup we bring here (static environment, primitive task and practically unlimited access to every piece of relevant information), it is difficult to induce a design principle, what are our chances of exposing biological design principles when more realistic conditions are examined?Implications to the way we do Biology are discussed.

View Article: PubMed Central - PubMed

Affiliation: Network Biology Research Laboratories, Technion - Israel Institute of Technology Haifa, Israel. marom@technion.ac.il

ABSTRACT
In this perspective we provide an example for the limits of reverse engineering in neuroscience. We demonstrate that application of reverse engineering to the study of the design principle of a functional neuro-system with a known mechanism, may result in a perfectly valid but wrong induction of the system's design principle. If in the very simple setup we bring here (static environment, primitive task and practically unlimited access to every piece of relevant information), it is difficult to induce a design principle, what are our chances of exposing biological design principles when more realistic conditions are examined? Implications to the way we do Biology are discussed.

No MeSH data available.