Search

Summary: Inductive generalization is adaptive in novel contexts for both biological and artificial intelligence. Spontaneous generalization in inexperienced animals raises questions on whether predispositions (evolutionarily acquired biases, or priors) enable generalization from sparse data, without reinforcement. We exposed neonate chicks to an artificial social partner of a specific color, and then looked at generalization on the red-yellow or blue-green ranges. Generalization was inconsistent with an unbiased model. Biases included asymmetrical generalization gradients, some preferences for unfamiliar stimuli, different speed of learning, faster learning for colors infrequent in the natural spectrum. Generalization was consistent with a Bayesian model that incorporates predispositions as initial preferences and treats the learning process as an update of predispositions. Newborn chicks are evolutionarily prepared for generalization, via biases independent from experience, reinforcement, or supervision. To solve the problem of induction, biological and artificial intelligence can use biases tuned to infrequent stimuli, such as the red and blue colors.

Summary: Varying levels of numerical cognition have been found in several animal species. Bees, in particular, have been argued to be able to count up to four items and solve complex numerical tasks. Here we present an exceedingly simple neural circuit that, when provided with the actual visual input that the bee is receiving while carrying out the task, can make reliable estimates on the number of items in the display. Thus we suggest that the elegance of numerical problem solving in bees might not lie in the formation of numerical concepts (such as “more,” “less,” or “zero”), but in the use of specific flight movements to scan targets, which streamlines the visual input and so renders the task of counting computationally inexpensive. Careful examination of the actual inspection strategies used by animals might reveal that animals often employ active scanning behaviors as shortcuts to simplify complex visual pattern discrimination tasks. : Neuroscience; Cognitive Neuroscience; Biocomputational Method Subject Areas: Neuroscience, Cognitive Neuroscience, Biocomputational Method

Organisms that can learn about their environment and modify their behaviour appropriately during their lifetime are more likely to survive and reproduce than organisms that do not. While associative learning - the ability to detect correlated features of the environment - has been studied extensively in nervous systems, where the underlying mechanisms are reasonably well understood, mechanisms within single cells that could allow associative learning have received little attention. Here, using in silico evolution of chemical networks, we show that there exists a diversity of remarkably simple and plausible chemical solutions to the associative learning problem, the simplest of which uses only one core chemical reaction. We then asked to what extent a linear combination of chemical concentrations in the network could approximate the ideal Bayesian posterior of an environment given the stimulus history so far? This Bayesian analysis revealed the 'memory traces' of the chemical network. The implication of this paper is that there is little reason to believe that a lack of suitable phenotypic variation would prevent associative learning from evolving in cell signalling, metabolic, gene regulatory, or a mixture of these networks in cells.

We propose a previously unrecognized kind of informational entity in the brain that is capable of acting as the basis for unlimited hereditary variation in neuronal networks. This unit is a path of activity through a network of neurons, analogous to a path taken through a hidden Markov model. To prove in principle the capabilities of this new kind of informational substrate, we show how a population of paths can be used as the hereditary material for a neuronally implemented genetic algorithm, (the swiss-army knife of black-box optimization techniques) which we have proposed elsewhere could operate at somatic timescales in the brain. We compare this to the same genetic algorithm that uses a standard 'genetic' informational substrate, i.e. non-overlapping discrete genotypes, on a range of optimization problems. A path evolution algorithm (PEA) is defined as any algorithm that implements natural selection of paths in a network substrate. A PEA is a previously unrecognized type of natural selection that is well suited for implementation by biological neuronal networks with structural plasticity. The important similarities and differences between a standard genetic algorithm and a PEA are considered. Whilst most experiments are conducted on an abstract network model, at the conclusion of the paper a slightly more realistic neuronal implementation of a PEA is outlined based on Izhikevich spiking neurons. Finally, experimental predictions are made for the identification of such informational paths in the brain.