Two-Dimensional Nonlinear Categorization by Caenorhabditis elegans

Author(s)

Parth DesaiFollow

School Name

Spring Valley High School

Grade Level

12th Grade

Presentation Topic

Zoology

Presentation Type

Non-Mentored

Abstract

Caenorhabditis elegans have been utilized to study the nervous system for decades. However, the computational prowess of the neural network of the microorganism is not fully understood. This experiment tested whether C. elegans are capable of learning two-dimensional categories that are linearly as well as non-linearly separable. It was hypothesized that C. elegans would be able to learn both linearly and nonlinearly separable classes. This was tested by using two dimensions of potassium chloride (KCI) and color, whose presence/absence was associated with the presence/absence of food (E. Coli). C. elegans were trained to learn one of three relations: AND and OR (both linear) and XOR (nonlinear). After training, learning was then tested by examining the movement of C. elegans in separate dishes. In a separate experiment, the reverse associations (absence/presence of food) representing the same functions were also trained and tested. The results showed that C. elegans can learn these positive and negative associations for both linear and nonlinear cases. This advances our understanding in two ways. First, it shows that the microorganism can learn two-dimensional problems, i.e., simultaneously considering two variables to determine behavior. Secondly, it shows for the first time, that their neural networks possess sufficient computational complexity to learn nonlinear boundaries. This will put them in a different class of organisms, capable of solving more complex problems than previously realized.For the AND Gate, X-squared = 83.42, p < 0.001. For the OR Gate 100/300 test, x-squared = 73.657, p < 0.001. For the NAND Gate 1.1454, p = 0.28445. For the NOR Gate, 9.79, p = 0.001755. For the XOR Gate, 49.278, p < 0.001.

Location

HSS 107

Start Date

4-2-2022 11:15 AM

Presentation Format

Oral and Written

Group Project

No

COinS
 
Apr 2nd, 11:15 AM

Two-Dimensional Nonlinear Categorization by Caenorhabditis elegans

HSS 107

Caenorhabditis elegans have been utilized to study the nervous system for decades. However, the computational prowess of the neural network of the microorganism is not fully understood. This experiment tested whether C. elegans are capable of learning two-dimensional categories that are linearly as well as non-linearly separable. It was hypothesized that C. elegans would be able to learn both linearly and nonlinearly separable classes. This was tested by using two dimensions of potassium chloride (KCI) and color, whose presence/absence was associated with the presence/absence of food (E. Coli). C. elegans were trained to learn one of three relations: AND and OR (both linear) and XOR (nonlinear). After training, learning was then tested by examining the movement of C. elegans in separate dishes. In a separate experiment, the reverse associations (absence/presence of food) representing the same functions were also trained and tested. The results showed that C. elegans can learn these positive and negative associations for both linear and nonlinear cases. This advances our understanding in two ways. First, it shows that the microorganism can learn two-dimensional problems, i.e., simultaneously considering two variables to determine behavior. Secondly, it shows for the first time, that their neural networks possess sufficient computational complexity to learn nonlinear boundaries. This will put them in a different class of organisms, capable of solving more complex problems than previously realized.For the AND Gate, X-squared = 83.42, p < 0.001. For the OR Gate 100/300 test, x-squared = 73.657, p < 0.001. For the NAND Gate 1.1454, p = 0.28445. For the NOR Gate, 9.79, p = 0.001755. For the XOR Gate, 49.278, p < 0.001.