![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
hello_robottoHi, CC! Long time, no see, busy with getting the Ph.D and all that. Hope you guys are all doing well and staying healthy and sane. I'm probably not back for good, but I still subscribe to several Creatures blogs in my RSS feed, and seeing the CCSF 2021 posts popping up in my feed made me feel nostalgic for these games. And I'm excited to see them coming to Steam! Crossing my fingers that we finally get an official C2 release that works on modern operating systems... maybe even one that fixes the bug that makes worlds inevitably crash at a certain age?!
I was even inspired to dig up and update this old mini-paper on how the Creatures C3/DS brains are set up; I originally wrote it for an independent study way back in 2014 or so. And while updating it, I promptly started a DS feral run inspired by it - one where all the starting creatures had their brain lobes set to be mutable. Not expecting anything particularly interesting to come of it, but you never know :)
Anyway, if you're curious about how the Creatures brains work, but find them difficult or intimidating to understand, hopefully this will help get you started. It can be hard to figure out a brain in a vat or what all those rules in the genetics kit are doing to them!
Sources
My main sources for this were:
The official Creatures 3 genetics kit
SeeYou7's German guide to the brain - unfortunately lost to time and not on the Internet Archive
This list of what some of the state variable rules mean
More explanation of the state variable rules
This recent and very helpful documentation of the state variable rules gleaned from the source code
This article written by Steve Grand; doesn't fully apply to C3/DS, but gives an idea of where it came from
Creation by Steve Grand – more background information and a few concepts; again, written with C1 in mind
Structure
Each creature's brain is made up of an artificial neural network, which is a data structure that has been used in artificial intelligence for decades. It is 'inspired by' the real brains we see in animals such as ourselves, and there are different variations on the concept, some of which are more 'inspired' and some of which are more abstracted. The version used in Creatures is unique as far as I have seen; if you study neural networks for AI, they are set up pretty differently.
Each neuron is essentially one node in the brain. It contains up to eight values: its state, its input, its output, 'susceptibility' and 'susceptibility input' (I'm not entirely sure what these do more specifically), 'neural growth factor', plus two additional values simply known as var 5 and var 6. These neurons are arranged into groups, called lobes, with different functions, described below. Neurons are connected to neurons in other lobes via dendrites, which are one-way connections that can have different behaviors; some simply transfer the state of one neuron to another, while others may do more complex work.
The creature's brain structure is governed by its genetics; this means that it can be edited in the genetics kit. In Creatures 3/DS, none of the brain structures are allowed to mutate in the default creatures. This is probably to reduce the number of still-born or severely disabled creatures. But theoretically creatures with new, perhaps better brain structures could be allowed to evolve by changing the mutation rates. (My experience, plus the lack of evolved super-smart creatures in C1/C2 where brains could mutate by default, suggests that this is not very likely in the Creatures system... but theoretically, it is possible.) The genetics also define a creature's instincts, which are processed in the brain during sleep. It is possible to have an instinctless creature, although they often die young if the user does not teach them how to survive.
The C3/DS brain has 15 lobes. Each lobe has state variable rules (SV rules from here on). A state variable is a variable that represents the state of a system at some point in time and can be used to predict the behavior of the system in the future. These rules govern how the neurons' values initialize and then update. In explaining some of these rules, I will refer to the accumulator; this is where math happens before values are saved to the neuron. You can think of it like the current value displayed in a calculator – you can do operations to it and multiply it and so on by saved values or new inputs, before copying the value somewhere else.
The update rules run every 4 ticks (under ideal circumstances, there should be 20 ticks per second at normal speed). This is one reason why having a ton of creatures in your world will slow it down – although the engine will try to offset the updates so not every creature is updating on the same ticks, there's a lot of updates happening several times a second, both in their brains and in their biochemistry. Some of the lobes also don't function while the creature is asleep.
With two exceptions, most of the lobes are connected to at least one other lobe. Even these two exceptions are useful for the creature, as they interact with the creature's biochemistry. For example: 'I see a creature of the opposite sex and same genus' → release pheromones.
The rest of this guide covers the individual lobes and their functions in the creature brain as best as I understand it. This includes both the default norns and the commonly used CFE norns; there are other genomes out there that have additional lobes or modified rules. The lobes are listed in the order they appear in the genetics kit when sorted by gene type, with the exception of the elevation lobe, which is new to CFE norns. Please note that the brain tracts/connections are also quite important - most of the CFE brain edits are related to the tracts rather than the lobes themselves - but they are not covered here. Maybe another time :)
Drive lobe
The neurons in this lobe store the intensity of the creature's drives – how hot, cold, angry, bored, tired, and so on it is. It contains 20 neurons.
The SV rules are quite simple:
store in | neuron | state
This writes the input value (in this case, the intensity of the given drive) to the particular neuron without doing anything to it.
The drive lobe outputs to the combination lobe, to the friends and foe lobe, and the mood lobe. It receives an input from the response lobe.
Decision lobe
This lobe decides the creature's next action. It contains 13 neurons.
The SV rules are as follows:
load from | neuron | input
store in | neuron | state
blank | neuron | input
blank | neuron | var 5
if < | spare neuron | state
stop
blank | spare neuron | var 5
store in | neuron | var 5
register as spare
This is somewhat more complicated. It essentially says this:
1. Load the value of the input neuron.
2. Write the value to the current neuron.
3. Blank the value of the input neuron.
4. Make the 5th state of the neuron blank.
5. If the current value of the neuron is less than the value of the 'spare neuron', which is really just one of the lobe's neurons, which may change, do the next line, else skip the next line.
6. Stop!
7. (Else) Blank the spare neuron's 5th state.
8. Store the current value in the current neuron's 5th state.
9. Register this neuron as the spare neuron.
This was known as the 'winner takes all' system in previous versions of Creatures. Essentially, the most active neuron will cause the others to become inactive. The creature's decision is decided by the most active neuron.
The decision lobe outputs to the response lobe and takes input from the combination and verb lobes.
Attention lobe
This lobe decides what the creature's attention is focused on. It contains 40 neurons.
The SV rules are as follows:
load from | neuron | input
no operation
no operation
store in | neuron | state
blank | neuron | input
if > | spare neuron | state
register as spare
This is another 'winner takes all' system, though without suppressing the activity of other neurons.
The attention lobe has no outputs and takes input from the combination lobe.
Vision lobe
This lobe has information on what the creature can see, how far (horizontally) the object is, and in what direction. It contains 40 neurons.
The SV rules are simple:
store in | neuron | state
Again, this simply stores a value in the neuron without doing anything to it.
The vision lobe outputs to the stimuli, movement, and smell lobes (this last is to keep either smells or vision from overwhelming the importance of the other) and has no input.
Movement lobe
This lobe tracks the relative velocity (change in distance) of the objects in vision. It contains 40 neurons.
The SV rules are as follows:
load from | neuron | input
multiply by | value | .9355
add | neuron | state
set tend rate | value | .0484
tend to | zero
store in | neuron | state
blank | neuron | input
This essentially says: load the accumulator with the value from the input neuron and multiply by .9355. Add to this accumulator the state of the neuron. Set the tend rate to .0484. The tend rate is used in the next operation: tend towards zero. That is to say: the NEW value of the accumulator = the CURRENT value * (1 – tend rate). Now, store the accumulator in the neuron and blank the input.
The movement lobe outputs to the stimuli lobe and takes input from the vision lobe.
Combination lobe
This lobe combines the output of the various sensory lobes. It is the largest and most complex lobe, with 440 neurons. Each neuron can have one, two, or three input dendrites, which can change if the connections are weak. This lobe essentially allows the creature to learn to recognize different situations ('I am angry' + 'I see another norn', for example) and how to react to them ('Hit', perhaps!).
The SV rules are as follows:
set tend rate | value | .8145
load from | neuron | input
tend to | neuron | state
store in | neuron | state
blank | neuron | input
stop if < | spare neuron | state
blank | spare neuron | output
store in | neuron | output
preserve neuron sv | value integer | 3
load from | value | .8024
if non-zero | chemical | REM
load from | one
if > | neuron | susceptibility
store in | neuron | susceptibility
restore spare neuron var | value integer | 3
register as spare
This is complicated! Here it is broken down a little:
1. Set the tend rate we covered just above to .8145
2. Load the accumulator with the neuron's input
3. This is a little tricky. This is like 'tend to zero', but it tends to the current state of the neuron instead. So the NEW accumulator value = CURRENT value + (tend rate * (neuron's state – CURRENT value). So the new value approaches the value of the neuron's state at a rate proportional to the difference between it and the current value, multiplied by the tend rate.
4. Write this accumulator value to the neuron.
5. Blank the neuron's input.
6. If this neuron is less than the spare neuron, stop doing anything.
7. Otherwise, blank the spare neuron's output.
8. Write the accumulator value to the neuron's output.
9. This sets the susceptibility input to the value from the neuron's own variable '3', which is its susceptibility (although I'm not entirely sure what this meaningfully does).
10. Change the accumulator to .8024.
11. If the chemical REM is not zero, do the next line. That is to say, if the creature is sleeping and processing its instincts....
12. Change the accumulator to 1.
13. If the accumulator is greater than the neuron's susceptibility, then:
14. Write the accumulator value to the neuron's susceptibility.
15. Set the space neuron's susceptibility to its own variable '3' (which, again, is its susceptibility).
16. Register this neuron as the spare neuron.
Incidentally, this lobe uses all the lines allowed in the SV rules editor – sixteen is the maximum amount of instructions.
To put this slightly more simply, the inputs to each neuron are added together and depending on the difference between the input and the neuron's current value, the current value is adjusted. If this neuron is not the 'winner', the neuron with the higher input, that is all for this update. If it is, I think what is going on is, the accumulator is changed to the default susceptibility value. Next, there is a check for whether the creature is dreaming. If so, the susceptibility value is made higher so that the instincts processing can more easily change the neuron's values. At the end, the neuron is registered as the spare neuron.
The combination lobe outputs to the attention and decision lobes, and takes input from the drive, stimuli, verb, and friends and foe lobes.
Stimuli lobe
This lobe keeps track of what is stimulating the creature's senses. It contains 40 neurons. Its SV rules are as follows:
load from | neuron | input
if < | zero
blank | accumulator
store in | neuron | state
blank | neuron | input
This is relatively simple – it stores whatever input it receives, as long as it is not negative.
The stimuli lobe outputs to the combination lobe, and receives input from the vision, movement, noun, and smell lobes. In CFE creatures, it also receives input from the elevation lobe.
Noun lobe
This lobe activates when the creature hears a noun being spoken, either by the user or by other creatures. It contains 40 neurons. Its SV rules are as follows:
no operation
multiply by | value | .6371
add | neuron | state
set tend rate | value | .0484
tend to | zero
store in | neuron | state
if > | value | .004
stop
load from | random
multiply by | value | .0403
add and store in neuron | state
Mostly self-explanatory: input is multiplied by .6371, added to the neuron's state, and the new value is relaxed toward zero. I'm not entirely sure what the significance of the last three lines is; notes in the genetics kit say 'wander behavior' next to it.
The noun lobe outputs to the stimuli lobe, and has no inputs.
Verb lobe
This lobe activates when the creature hears a verb being spoken, either by the user or by other creatures. It contains 13 neurons. Its SV rules are identical to those of the noun lobe.
The verb lobe outputs to the combination and decision lobes, and has no inputs.
Detail lobe
This lobe registers extra information about the object being viewed – 'it is being carried by me', 'it is a creature', and so on. It contains 16 neurons. The SV rules are very simple:
store in | neuron | state
The detail lobe has no inputs or outputs.
Situation lobe
This lobe registers extra information about the creature itself - 'I am in a vehicle', 'I am falling', 'I am carrying something', and so on. It contains 16 neurons. The SV rules are very simple:
store in | neuron | state
The situation lobe has no inputs or outputs.
Response lobe
This lobe is affected by different stimulus genes. It contains 20 neurons. Again, the SV rules are very simple:
store in | neuron | state
The response lobe outputs to the drive lobe and receives input from the decision lobe.
Smell lobe
This lobe registers what scent a creature can smell, as well as how strong the scent is. It contains 40 neurons. Again, the SV rules are very simple:
store in | neuron | state
The smell lobe outputs to the stimuli lobe, and has input from the vision lobe.
Friends and foe lobe
This lobe contains the creature's feelings, positive or negative, about other creatures it knows. It contains 36 neurons. The SV rules are as follows:
load from | zero
set tend rate | value | .8024
tend to, store in | neuron | susceptibility
no operation
load from | neuron | state
add | neuron | var 6
store in | neuron | output
blank | neuron | input
Essentially, it decays the neuron's learning susceptibility before affecting the neuron's output. The engine is also involved in the working of this lobe (see the command FORF).
However, in the default game, this... doesn't work. Creatures don't make friends or foes! Judging by the required edits to make a CFE creature, this appears to be mostly an issue of things not being connected up right, as the edits deal with editing brain tracts (connections) rather than fixing up the ForF lobe itself. There is also a new lobe for the CFE creatures, though, called the ForF support lobe, that has 36 neurons and no SV rules.
The friends and foe lobe outputs to the combination lobe, and has input from the drive and mood lobes. In the CFE norns, it also has both input from and output to the ForF support lobe.
Mood lobe
This lobe decides the overall mood of the creature – if it is fearful, or angry, or so on. It contains only one neuron. The SV rules are as follows:
load from | neuron | var 5
add | neuron | var 6
store in | neuron | susceptibility
load from | neuron | state
set tend rate | value | .2016
tend to | neuron | susceptibility
store in | neuron | state
no operation
no operation
no operation
no operation
no operation
store in | neuron | output
blank | neuron | var 5
blank | neuron | var 6
stop
The neuron's fifth and sixth variables are added together to make the susceptibility, and the neuron's state is then tended toward this susceptibility.I'm not entirely sure what the five lines of 'no operation' means; there are comments by these lines saying 'load chemical' and 'alter'. Possibly the engine is doing something during these lines? Apparently, nothing happens during these lines; they may be left over from unimplemented ideas during development.
The mood lobe outputs to the friends and foe lobe and takes input from the drive lobe.
Elevation lobe
The elevation lobe is a new lobe for CFE creatures that helps them judge how far away an object is vertically (and thus if it's in reach or not) - the vision lobe does this for horizontal distance. Like the vision lobe, it has 40 neurons. The SV rules are simple:
store in | neuron | state
While this is added in the CFE edits, the wiki article implies that the code to calculate elevation was already present in C3 (but broken), and perhaps came with DS (but wasn't implemented?).
The elevation lobe has no inputs and outputs to the stimuli lobe.
I was even inspired to dig up and update this old mini-paper on how the Creatures C3/DS brains are set up; I originally wrote it for an independent study way back in 2014 or so. And while updating it, I promptly started a DS feral run inspired by it - one where all the starting creatures had their brain lobes set to be mutable. Not expecting anything particularly interesting to come of it, but you never know :)
Anyway, if you're curious about how the Creatures brains work, but find them difficult or intimidating to understand, hopefully this will help get you started. It can be hard to figure out a brain in a vat or what all those rules in the genetics kit are doing to them!
Sources
My main sources for this were:
The official Creatures 3 genetics kit
SeeYou7's German guide to the brain - unfortunately lost to time and not on the Internet Archive
This list of what some of the state variable rules mean
More explanation of the state variable rules
This recent and very helpful documentation of the state variable rules gleaned from the source code
This article written by Steve Grand; doesn't fully apply to C3/DS, but gives an idea of where it came from
Creation by Steve Grand – more background information and a few concepts; again, written with C1 in mind
Structure
Each creature's brain is made up of an artificial neural network, which is a data structure that has been used in artificial intelligence for decades. It is 'inspired by' the real brains we see in animals such as ourselves, and there are different variations on the concept, some of which are more 'inspired' and some of which are more abstracted. The version used in Creatures is unique as far as I have seen; if you study neural networks for AI, they are set up pretty differently.
Each neuron is essentially one node in the brain. It contains up to eight values: its state, its input, its output, 'susceptibility' and 'susceptibility input' (I'm not entirely sure what these do more specifically), 'neural growth factor', plus two additional values simply known as var 5 and var 6. These neurons are arranged into groups, called lobes, with different functions, described below. Neurons are connected to neurons in other lobes via dendrites, which are one-way connections that can have different behaviors; some simply transfer the state of one neuron to another, while others may do more complex work.
The creature's brain structure is governed by its genetics; this means that it can be edited in the genetics kit. In Creatures 3/DS, none of the brain structures are allowed to mutate in the default creatures. This is probably to reduce the number of still-born or severely disabled creatures. But theoretically creatures with new, perhaps better brain structures could be allowed to evolve by changing the mutation rates. (My experience, plus the lack of evolved super-smart creatures in C1/C2 where brains could mutate by default, suggests that this is not very likely in the Creatures system... but theoretically, it is possible.) The genetics also define a creature's instincts, which are processed in the brain during sleep. It is possible to have an instinctless creature, although they often die young if the user does not teach them how to survive.
The C3/DS brain has 15 lobes. Each lobe has state variable rules (SV rules from here on). A state variable is a variable that represents the state of a system at some point in time and can be used to predict the behavior of the system in the future. These rules govern how the neurons' values initialize and then update. In explaining some of these rules, I will refer to the accumulator; this is where math happens before values are saved to the neuron. You can think of it like the current value displayed in a calculator – you can do operations to it and multiply it and so on by saved values or new inputs, before copying the value somewhere else.
The update rules run every 4 ticks (under ideal circumstances, there should be 20 ticks per second at normal speed). This is one reason why having a ton of creatures in your world will slow it down – although the engine will try to offset the updates so not every creature is updating on the same ticks, there's a lot of updates happening several times a second, both in their brains and in their biochemistry. Some of the lobes also don't function while the creature is asleep.
With two exceptions, most of the lobes are connected to at least one other lobe. Even these two exceptions are useful for the creature, as they interact with the creature's biochemistry. For example: 'I see a creature of the opposite sex and same genus' → release pheromones.
The rest of this guide covers the individual lobes and their functions in the creature brain as best as I understand it. This includes both the default norns and the commonly used CFE norns; there are other genomes out there that have additional lobes or modified rules. The lobes are listed in the order they appear in the genetics kit when sorted by gene type, with the exception of the elevation lobe, which is new to CFE norns. Please note that the brain tracts/connections are also quite important - most of the CFE brain edits are related to the tracts rather than the lobes themselves - but they are not covered here. Maybe another time :)
Drive lobe
The neurons in this lobe store the intensity of the creature's drives – how hot, cold, angry, bored, tired, and so on it is. It contains 20 neurons.
The SV rules are quite simple:
store in | neuron | state
This writes the input value (in this case, the intensity of the given drive) to the particular neuron without doing anything to it.
The drive lobe outputs to the combination lobe, to the friends and foe lobe, and the mood lobe. It receives an input from the response lobe.
Decision lobe
This lobe decides the creature's next action. It contains 13 neurons.
The SV rules are as follows:
load from | neuron | input
store in | neuron | state
blank | neuron | input
blank | neuron | var 5
if < | spare neuron | state
stop
blank | spare neuron | var 5
store in | neuron | var 5
register as spare
This is somewhat more complicated. It essentially says this:
1. Load the value of the input neuron.
2. Write the value to the current neuron.
3. Blank the value of the input neuron.
4. Make the 5th state of the neuron blank.
5. If the current value of the neuron is less than the value of the 'spare neuron', which is really just one of the lobe's neurons, which may change, do the next line, else skip the next line.
6. Stop!
7. (Else) Blank the spare neuron's 5th state.
8. Store the current value in the current neuron's 5th state.
9. Register this neuron as the spare neuron.
This was known as the 'winner takes all' system in previous versions of Creatures. Essentially, the most active neuron will cause the others to become inactive. The creature's decision is decided by the most active neuron.
The decision lobe outputs to the response lobe and takes input from the combination and verb lobes.
Attention lobe
This lobe decides what the creature's attention is focused on. It contains 40 neurons.
The SV rules are as follows:
load from | neuron | input
no operation
no operation
store in | neuron | state
blank | neuron | input
if > | spare neuron | state
register as spare
This is another 'winner takes all' system, though without suppressing the activity of other neurons.
The attention lobe has no outputs and takes input from the combination lobe.
Vision lobe
This lobe has information on what the creature can see, how far (horizontally) the object is, and in what direction. It contains 40 neurons.
The SV rules are simple:
store in | neuron | state
Again, this simply stores a value in the neuron without doing anything to it.
The vision lobe outputs to the stimuli, movement, and smell lobes (this last is to keep either smells or vision from overwhelming the importance of the other) and has no input.
Movement lobe
This lobe tracks the relative velocity (change in distance) of the objects in vision. It contains 40 neurons.
The SV rules are as follows:
load from | neuron | input
multiply by | value | .9355
add | neuron | state
set tend rate | value | .0484
tend to | zero
store in | neuron | state
blank | neuron | input
This essentially says: load the accumulator with the value from the input neuron and multiply by .9355. Add to this accumulator the state of the neuron. Set the tend rate to .0484. The tend rate is used in the next operation: tend towards zero. That is to say: the NEW value of the accumulator = the CURRENT value * (1 – tend rate). Now, store the accumulator in the neuron and blank the input.
The movement lobe outputs to the stimuli lobe and takes input from the vision lobe.
Combination lobe
This lobe combines the output of the various sensory lobes. It is the largest and most complex lobe, with 440 neurons. Each neuron can have one, two, or three input dendrites, which can change if the connections are weak. This lobe essentially allows the creature to learn to recognize different situations ('I am angry' + 'I see another norn', for example) and how to react to them ('Hit', perhaps!).
The SV rules are as follows:
set tend rate | value | .8145
load from | neuron | input
tend to | neuron | state
store in | neuron | state
blank | neuron | input
stop if < | spare neuron | state
blank | spare neuron | output
store in | neuron | output
preserve neuron sv | value integer | 3
load from | value | .8024
if non-zero | chemical | REM
load from | one
if > | neuron | susceptibility
store in | neuron | susceptibility
restore spare neuron var | value integer | 3
register as spare
This is complicated! Here it is broken down a little:
1. Set the tend rate we covered just above to .8145
2. Load the accumulator with the neuron's input
3. This is a little tricky. This is like 'tend to zero', but it tends to the current state of the neuron instead. So the NEW accumulator value = CURRENT value + (tend rate * (neuron's state – CURRENT value). So the new value approaches the value of the neuron's state at a rate proportional to the difference between it and the current value, multiplied by the tend rate.
4. Write this accumulator value to the neuron.
5. Blank the neuron's input.
6. If this neuron is less than the spare neuron, stop doing anything.
7. Otherwise, blank the spare neuron's output.
8. Write the accumulator value to the neuron's output.
9. This sets the susceptibility input to the value from the neuron's own variable '3', which is its susceptibility (although I'm not entirely sure what this meaningfully does).
10. Change the accumulator to .8024.
11. If the chemical REM is not zero, do the next line. That is to say, if the creature is sleeping and processing its instincts....
12. Change the accumulator to 1.
13. If the accumulator is greater than the neuron's susceptibility, then:
14. Write the accumulator value to the neuron's susceptibility.
15. Set the space neuron's susceptibility to its own variable '3' (which, again, is its susceptibility).
16. Register this neuron as the spare neuron.
Incidentally, this lobe uses all the lines allowed in the SV rules editor – sixteen is the maximum amount of instructions.
To put this slightly more simply, the inputs to each neuron are added together and depending on the difference between the input and the neuron's current value, the current value is adjusted. If this neuron is not the 'winner', the neuron with the higher input, that is all for this update. If it is, I think what is going on is, the accumulator is changed to the default susceptibility value. Next, there is a check for whether the creature is dreaming. If so, the susceptibility value is made higher so that the instincts processing can more easily change the neuron's values. At the end, the neuron is registered as the spare neuron.
The combination lobe outputs to the attention and decision lobes, and takes input from the drive, stimuli, verb, and friends and foe lobes.
Stimuli lobe
This lobe keeps track of what is stimulating the creature's senses. It contains 40 neurons. Its SV rules are as follows:
load from | neuron | input
if < | zero
blank | accumulator
store in | neuron | state
blank | neuron | input
This is relatively simple – it stores whatever input it receives, as long as it is not negative.
The stimuli lobe outputs to the combination lobe, and receives input from the vision, movement, noun, and smell lobes. In CFE creatures, it also receives input from the elevation lobe.
Noun lobe
This lobe activates when the creature hears a noun being spoken, either by the user or by other creatures. It contains 40 neurons. Its SV rules are as follows:
no operation
multiply by | value | .6371
add | neuron | state
set tend rate | value | .0484
tend to | zero
store in | neuron | state
if > | value | .004
stop
load from | random
multiply by | value | .0403
add and store in neuron | state
Mostly self-explanatory: input is multiplied by .6371, added to the neuron's state, and the new value is relaxed toward zero. I'm not entirely sure what the significance of the last three lines is; notes in the genetics kit say 'wander behavior' next to it.
The noun lobe outputs to the stimuli lobe, and has no inputs.
Verb lobe
This lobe activates when the creature hears a verb being spoken, either by the user or by other creatures. It contains 13 neurons. Its SV rules are identical to those of the noun lobe.
The verb lobe outputs to the combination and decision lobes, and has no inputs.
Detail lobe
This lobe registers extra information about the object being viewed – 'it is being carried by me', 'it is a creature', and so on. It contains 16 neurons. The SV rules are very simple:
store in | neuron | state
The detail lobe has no inputs or outputs.
Situation lobe
This lobe registers extra information about the creature itself - 'I am in a vehicle', 'I am falling', 'I am carrying something', and so on. It contains 16 neurons. The SV rules are very simple:
store in | neuron | state
The situation lobe has no inputs or outputs.
Response lobe
This lobe is affected by different stimulus genes. It contains 20 neurons. Again, the SV rules are very simple:
store in | neuron | state
The response lobe outputs to the drive lobe and receives input from the decision lobe.
Smell lobe
This lobe registers what scent a creature can smell, as well as how strong the scent is. It contains 40 neurons. Again, the SV rules are very simple:
store in | neuron | state
The smell lobe outputs to the stimuli lobe, and has input from the vision lobe.
Friends and foe lobe
This lobe contains the creature's feelings, positive or negative, about other creatures it knows. It contains 36 neurons. The SV rules are as follows:
load from | zero
set tend rate | value | .8024
tend to, store in | neuron | susceptibility
no operation
load from | neuron | state
add | neuron | var 6
store in | neuron | output
blank | neuron | input
Essentially, it decays the neuron's learning susceptibility before affecting the neuron's output. The engine is also involved in the working of this lobe (see the command FORF).
However, in the default game, this... doesn't work. Creatures don't make friends or foes! Judging by the required edits to make a CFE creature, this appears to be mostly an issue of things not being connected up right, as the edits deal with editing brain tracts (connections) rather than fixing up the ForF lobe itself. There is also a new lobe for the CFE creatures, though, called the ForF support lobe, that has 36 neurons and no SV rules.
The friends and foe lobe outputs to the combination lobe, and has input from the drive and mood lobes. In the CFE norns, it also has both input from and output to the ForF support lobe.
Mood lobe
This lobe decides the overall mood of the creature – if it is fearful, or angry, or so on. It contains only one neuron. The SV rules are as follows:
load from | neuron | var 5
add | neuron | var 6
store in | neuron | susceptibility
load from | neuron | state
set tend rate | value | .2016
tend to | neuron | susceptibility
store in | neuron | state
no operation
no operation
no operation
no operation
no operation
store in | neuron | output
blank | neuron | var 5
blank | neuron | var 6
stop
The neuron's fifth and sixth variables are added together to make the susceptibility, and the neuron's state is then tended toward this susceptibility.
The mood lobe outputs to the friends and foe lobe and takes input from the drive lobe.
Elevation lobe
The elevation lobe is a new lobe for CFE creatures that helps them judge how far away an object is vertically (and thus if it's in reach or not) - the vision lobe does this for horizontal distance. Like the vision lobe, it has 40 neurons. The SV rules are simple:
store in | neuron | state
While this is added in the CFE edits, the wiki article implies that the code to calculate elevation was already present in C3 (but broken), and perhaps came with DS (but wasn't implemented?).
The elevation lobe has no inputs and outputs to the stimuli lobe.
