Tissue-Level Intelligence �via Synthetic Living Constructs

Wesley Clawson

​

Levin Lab

1

WCDO 2025

Burlington, VT

Tissue-Level Intelligence �via Synthetic Living Constructs

2

WCDO 2025

Burlington, VT

COMPUTATIONALLY DESIGNED

Wesley Clawson

​

Levin Lab

Library of Intelligence Tests

Catalogue of “bio-rithms”

4

5

wow so smart

6

wow so smart

any biological system

7

“constituted by a more or less broad series of elements [called] 'carriers of significance' or 'marks' which are the only things that interest the animal system“

• Biological systems have a remarkable ability to adapt, or learn, to a variety of change

​

• Importantly, this learning can occur across a variety of scales and spaces

​

8

Biological Learning

Figures adapted from Levin (2021), with permission

• Biological systems have a remarkable ability to adapt, or learn, to a variety of changes

​

• Importantly, this learning can occur across a variety of scales and spaces

​

• Intelligence is the degree of competency of navigating any space

​

9

Intelligence

Figures adapted from Levin (2021), with permission

10

Biological Intelligence

Figures adapted from Levin (2021), with permission

11

Biological Intelligence

Behavior 2

Behavior 1

Marder, E., & Goaillard, J. M. (2006). Variability, compensation and homeostasis in neuron and network function. Nat Rev Neurosci, 7(7), 563-574.

Pyloric Rhythm

Marder, E., & Goaillard, J. M. (2006). Variability, compensation and homeostasis in neuron and network function. Nat Rev Neurosci, 7(7), 563-574.

Biological Computation

13

Adapted from Krakauer et al, 2017

Biological Computation

14

Adapted from Krakauer et al, 2017

Biological Computation

15

Adapted from Krakauer et al, 2017

Biological Computation

16

Adapted from Krakauer et al, 2017

Biological Computation

17

​

• Biological systems have a remarkable ability to adapt, or learn, to a variety of changes

​

• Importantly, this learning can occur across a variety of scales and spaces

​

• Intelligence is the degree of competency of navigating any space

​

• Multiscale competency – each scale is agential, or pretty darn good at its problem set

18

Biological Learning

Figures adapted from Levin (2021), with permission

Some ‘Mechanisms’ of Problem Solving

​

• Biological systems have a remarkable ability to adapt, or learn, to a variety of changes

​

• Importantly, this learning can occur across a variety of scales and spaces

​

• Intelligence is the degree of competency of navigating any space

​

• Multiscale competency – each scale is agential, or pretty darn good at its problem set

• Fundamental principles of multicellular plasticity are obscured as biology is evolved and complex�
• We propose teleonomic (goal-seeking, problem-solving) behavior in diverse problem spaces as a powerful invariant across possible beings regardless of composition or origin.

​

• Analysis of living forms that have never existed before is necessary to reveal deep design principles of life as it can be.

19

• How trainable/persuadable is a given system?
• Depends on system and scale
• Guess and check

​

​

​

20

Training Biology

Axis of Persuadability

Technological Approach to Mind Everywhere, Levin (2021)

• How trainable/persuadable is a given system?
• Depends on system and scale
• Guess and check

​

• We’re starting from C
• Train an artificial ‘why’
• Observe the ‘what’ and ‘how’ during and after training

​

​

21

Training Biology

Axis of Persuadability

why

how

what

Technological Approach to Mind Everywhere, Levin (2021)

22

wow so smart

any biological system

“constituted by a more or less broad series of elements [called] 'carriers of significance' or 'marks' which are the only things that interest the animal system“

23

Embodiment

an environment that acts on an organism

a mind that at least partially processes this input

a body that can change the relationship to environment

24

Virtual Embodiment

an environment that acts on an organism

a mind that at least partially processes this input

a body that can change the relationship to environment

25

Experimental Paradigm

26

This slide adapted with permission from Maxwell Bio

Measuring tissue activity

CMOS microelectrode array (MEA) allows for high-resolution recording of local field potential (LFP) and spiking activity

27

Local Field Potentials

Microelectrode array (MEA) allows for high-resolution recording of local field potential (LFP) and spiking activity

​

Obien M. et al., Front. In Neurosc. (2015)

28

Each square is an electrode

Each line is electrical activity

Each electrode is 18 μm

​

​

4 visible neurons in this area

1. Blue
2. Purple
3. Orange
4. Yellow

​

​

​

​

29

This is day 7 of growth

• You can see soma (negative deflection)
• You can see early dendritic growth (positive deflections)

30

If I highlight two neurons (blue and purple) you can see how the dendritic outreach (blue positive deflections) is heading towards the soma, and assumedly the axon, of the purple neuron

31

~ 3-5 ms

32

mind + behavior

E18 cortical neurons

BrainBits

Slow, easy to see, and culture wide

HDMEA

34

Learning in cultures

moving into the virtual world

Nathan Wu - Yale

Viraj Chhajed - UCLA

Wes Clawson - Tufts

Trevor Sullivan - Tufts

Evan Kolberg – John F Kennedy

36

graphic by: Alexis Pietak

Experimental Paradigm

Closing the loop

Holistic

Train the whole tissue��

Agential

Collective of active agents (neurons)�

Learning

train tissue to test intelligence

​

Closing the loop

Holistic

Train the whole tissue��

Agential

Collective of active agents (neurons)�

Learning

train tissue to test intelligence

​

H

A

L

https://project-hal.github.io/

Friston, K. The free-energy principle: a unified brain theory? (2010)

Training with H.A.L.

No stim

Random

Stim

Friston, K. The free-energy principle: a unified brain theory? (2010)

Training with H.A.L.

Friston, K. The free-energy principle: a unified brain theory? (2010)

Training with H.A.L.

No stim

Random

Stim

Rightwards Burst

Leftwards Burst

Friston, K. The free-energy principle: a unified brain theory? (2010)

Training with H.A.L.

Rightwards Burst

Leftwards Burst

No stim

Random

Stim

Friston, K. The free-energy principle: a unified brain theory? (2010)

Training with H.A.L.

No stim

Random

Stim

some current/future work

some current/future work

• What communication protocols are ‘best’ for task acquisition?

​

• What tissues work best in what capacity?

​

• How is learned behavior shared across a collective of collectives?

​

​

Stim B if A does X

Bringing CDOs and ‘HAL’ together

This is/could be a CDO in the future

Using CD-HAL to understand biology and biological computation

Using CD-HAL as biological compute

Bringing CDOs and ‘HAL’ together

• Environment
• “What perturbations produce xyz”
• “Is training signal A better than B?”

​

​

• Body
• “What would be good for task 4?”
• “How best to combine behavior I and J?”

​

• “Given this data, can you find systematic use of a given “?

​

​

how

what

why

Library of Intelligence Tests

Catalogue of “bio-rithms”

CD

CD

CD

CD

Thank you�

52

AFOSR Grant FA9550-22-1-0465 DARPA-EA-24-01-04-039

[your name here]

​

wesley.clawson@tufts.edu

Team Members:

• Caitlin Grasso
• Sydney Palmer
• Serena Meng

​

Past Team Members:

• Nathan Wu
• Trevor Sullivan
• Viraj Chhajed
• Evan Kolberg

Levin Lab Members:

• Devon Davidian
• Juanita Matthews
• Hamid Sediqi
• Axel De Baat
• Hananel Hazan
• Franz Kuchling

​

Levin Lab Admin:

• Bryanna Vacanti
• Emma Taddeo
• Robert Brucker

​