Sentience

Steve Jones, March 21, 2020

What is Sentience?

Sentience is described in the dictionary as "a sentient quality or state", and Sentient as "responsive to or conscious of sense impressions". Originating from the Latin sentientem, it was first used in the 1630s for the concept of an ability to feel.

A closely related term, Consciousness is described in Wikipedia to be "the sentience or awareness of internal or external existence". Oxford defines it to be "aware of and responding to one's surroundings; awake." This term has a more philosophical bent, since its usage to describe the more abstract ideas of thought has been much more prevalent than the former. To avoid philosophy, and define a problem that can be studied more concretely with science, we'll use the term sentience.

We needn't concern ourselves with the subjective experience of a fly, or squirrel, or whether they might feel emotions or be artistic. Our work doesn't include anthropomorphizing our subjects of study. It's actually miraculous enough to see that a squirrel can make its way in the world for a lifetime-- through changing seasons, finding food, avoiding predators, and engaging in social behavior. The squirrel is a great example of an organism with a highly-functioning central nervous system that enables it to survive in the world through sensing the environment and taking action with behavior. The mechanism that makes this possible: a brain, with less than 500,000,000 neurons.[1]

Not all animals have brains, yet all animals with brains exhibit behavior. The cilia of some bacteria allow them to move through liquid and seek out some chemistry over others, yet this is not interesting to us, nor are the many multicellular animals that have no nervous system yet manage to make their way in the world.

Plants are a different story. They are sessants, non-ambulatory living beings that have a fixed location, and don't have a need to move to find food or evade predators; instead, they respond to external stimuli (chemicals, light, and damage) on a slower timescale, still using cellular signaling without a nervous system. The study of this can be interesting, but is separate from our work.

Throughout our materials, we use the term Sentient to describe the entities with brains that allow them to (1) sense their environment and (2) exhibit meaningful and relevant behavior responsive to their environment. See the site Glossary for more terms.

It's important to understand that sentience is something that arises from a general purpose architecture, not from a fixed-function neuronal circuit like the reflex-arc or the electrical wiring of a home.

How Can We Discover How Sentience Arises?

The traditional way to learn about how brains work is to study their anatomy from the largest gross anatomical level to the cellular level and below to the chemistry level. Gordon Shepherd's book, The Synaptic Organization of the Brain[2] is an example of this kind of study.

In addition, scientific research with medical patients having brain lesions or disease has enabled identification of components of the brain which are involved in functional areas. These discoveries have guided brain research for decades. These studies can be very helpful to narrow our search for salient attributes.

Functional MRI similarly provides a monitoring tool that can be used to identify what areas of the brain are active when performing a task. These studies can also narrow our search for salient attributes.

High-Fidelity Mapping and Simulation of Natural Brains

Recent investigations by large research organizations have been able to produce whole- or partial-brain connectome atlases of organisms such as mice and humans. These results reveal both large-scale and small-scale complexities. At the higher levels, trunks between regions of the cortex can be seen, so that major interaction pathways are suggested. At the lower levels, canonical microcircuits forming minicolumns and macrocells in cortex are suggested. These findings can steer hypotheses in the direction of using these general paradigms in synthetic models.

These efforts focus on creating digital models of natural brains with high fidelity, especially gene expression, the physical structures arising from gene expression, and chemistry. The work a simulator must perform in order to simulate a single neuron's operation at this level is massive, because of the huge number of low-level details associated with that neuron. Multiply that by 87 billion, and then do that work every millisecond (the timescale on which action potential spiking occurs in natural brains), and it becomes a job for the biggest supercomputers previously used only for nuclear simulations, and weather forecasting. Even with this level of computing power, minutes or hours of compute time produce predictions for how a piece of cortex would operate for only a second at the cellular and chemistry levels. These simulations offer insights into the cellular infrastructure works that nature used to build brains, but they do not help answer how sentience arises-- that's a higher layer of abstraction.

Real-Time Functional Simulation of Synthetic Brains

A new way to understand how sentience arises is to create much simpler synthetic brain models that are inspired by biology, but which do not exist in nature. Synthetic brains, employing models designed by humans rather than evolved by nature, can be simulated in real-time, and can be connected to the real-world environment, so that they can sense the environment and engage it with behavior. This is a closed-loop system, because as behavior is elicited, it can cause the sensation of the environment, through vision, auditory, or tactile senses, to change.

Simulation of large models, as we've seen, can saturate even the largest computation resources, while still not not achieving real-time results. By reducing the fidelity of the simulation, a more functional model can be arrived at-- one which doesn't model the living substrate that nature uses to build brains, but instead models the functional behavior of the neurons themselves. The less fidelity required, the more performance is gained by the simulation, allowing it to run in real-time.

In his book The Singularity Is Near published in 2005, futurist Ray Kurzweil predicted that by the year 2045, computers would have sufficient storage capacity and processing power to simulate a human brain in real-time. It is interesting to note that even today, we are very close to achieving Kurzweil's prediction; NeuroSynthetica simulators running on high-end commercially-available PC hardware have run simple models exceeding 1GN (one "giganode", abbreviated GN) networks of 1bn neurons, each with thousands of connections to other nodes.

The Science of Synthetic Sentience

Today, using NeuroSynthetica's Workbench interactive modeling system, and the NeuroSynthetica Sentience Engine™ to host their simulations, researchers can investigate sentience using the scientific method that works so well in other disciplines-- hypothesize and test. An hypothesis is formed about how a computing fabric architecture that uses a replicated canonical circuit with plasticity might allow for closed-loop robotic motion. Within the graphical environment, the researcher uses the NeuroSynthetica SOMA™ modeling language to describe the canonical circuit and the object array, and the connections to input and output devices. The model is compiled quickly and the compiled object is generated on the simulation servers. Then the hypothesis is tested by running the simulation in real time, while it is connected to its environmental I/O devices. The behavior is observed, knowledge gained, and the model is refined. Through an iterative process, the researcher can quickly vary parameters associated with signals, receptors, and neurons, change the canonical circuit, or change the shape of the fabric, among many other possibilities.

Achieving sentience in real-time synthetic brain simulation is a collaborative effort. NeuroSynthetica supplies the technology, products, and services that make real-time functional simulations of models possible. Our community of researchers build models that ultimately exhibit behavior. Researchers around the world are invited to Join the Community, and gain access to NeuroSynthetica's technologies and products, and community resources on the online community portal. Already a community member? Click the Portal button.

Intrigued? Read the next article and learn more about Synthetic Brain Simulation.


Sources:


Copyright (C) 2021 NeuroSynthetica, LLC. All rights reserved. NeuroSynthetica™, the NeuroSynthetica logo, SOMA™ and Sentience Engine™ are trademarks or registered trademarks of NeuroSynthetica, LLC.