Jeff
Hawkins - A Thousand Brains |
|
|
Book |
Page |
|
Topic |
|
|
Hawkins - A Thousand Brains |
40 |
|
The neocortex
learns a predictive
model of the world. |
|
|
Hawkins - A Thousand Brains |
41 |
|
Predictions occur inside neurons. |
|
1 |
Hawkins - A Thousand Brains |
42 |
|
If you know
the melody, then your brain continually
predicts the next note. |
|
1 |
Hawkins - A Thousand Brains |
42 |
|
Predicting the next note in a melody is known as sequence memory. |
|
0 |
Hawkins - A Thousand Brains |
42 |
|
Sequence
memory is also used in language. Recognizing a spoken word is like recognizing a short melody. |
|
0 |
Hawkins - A Thousand Brains |
42 |
|
A word is defined by a sequence of
phonemes, whereas a melody is defined by a sequence of
musical intervals. |
|
0 |
Hawkins - A Thousand Brains |
43 |
|
Neurons have to figure out how much
context is necessary to make the right prediction. |
|
1 |
Hawkins -
A Thousand Brains |
44 |
|
Neurons have thousands, sometimes tens of thousands, of synapses, spaced along the branches of the dendrites. |
|
1 |
Hawkins - A Thousand Brains |
44 |
|
Some of the
dendrites are near the cell body, and some
dendrites are farther
away. |
|
0 |
Hawkins - A Thousand Brains |
44 |
|
How small and tightly packed the synapses are on dendrites. |
|
0 |
Hawkins - A Thousand Brains |
44 |
|
If proximal
synapses near the cell
body receive enough
input, they will fire. |
|
0 |
Hawkins - A Thousand Brains |
45 |
|
Less than 10% of the cell's synapses are in the proximal area near the
cell body. The other 90% are too far away to cause a spike. |
|
1 |
Hawkins - A Thousand Brains |
45 |
|
If an input
arrives at one of these distal synapses, it has almost no effect on the cell body. |
|
0 |
Hawkins - A Thousand Brains |
45 |
|
Once a dendrite
spike is activated, it travels along the dendrite until it reaches the cell body. |
|
0 |
Hawkins - A Thousand Brains |
45 |
|
Many
thousands of synapses found on the dendrites. |
|
0 |
Hawkins - A Thousand Brains |
45 |
|
Dendrites
spikes are predictions. |
|
0 |
Hawkins - A Thousand Brains |
46 |
|
In each minicolumn, multiple neurons respond to the same
input pattern. |
|
1 |
Hawkins - A Thousand Brains |
46 |
|
When an input
arrives that is unexpected, multiple neurons fire at once. |
|
0 |
Hawkins - A Thousand Brains |
46 |
|
If the input is predicted, then only the predictive state neurons become active. |
|
0 |
Hawkins - A Thousand Brains |
46 |
|
Unexpected
inputs
cause a lot more activity than expected ones. |
|
0 |
Hawkins - A Thousand Brains |
46 |
|
Prediction is a ubiquitous function of the
brain. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
Most predictions occur inside neurons. |
|
1 |
Hawkins - A Thousand Brains |
47 |
|
A prediction occurs when a neuron recognizes a pattern [in the ~10,000 synapses in its dendrites], creates a dendrite
spike, and is primed to
spike earlier than other neurons. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
Each neuron can recognize hundreds of patterns that predict when the neuron should become active. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
Prediction is built into the fabric of the neocortex, the neuron. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
We spent over a year testing the new neuron model and
sequence memory circuit. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
We wrote software
simulations that tested its capacity and were
surprised to find that as few as 20,000 neurons can learn thousands of complete
sequences |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
Sequence
memory continued to work even if 30% of the neurons died or if the input was noisy. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
We gained more confidence that our theory truly captured
what was happening in the neocortex. |
|
0 |
Hawkins - A Thousand Brains |
47 |
|
The function
of the cortical column
is reference frames. |
|
0 |
Hawkins - A Thousand Brains |
48 |
|
We could not
see a way to solve the problem, so we put the this question aside and worked on other things for a while. |
|
1 |
Hawkins - A Thousand Brains |
49 |
|
The movement
related signal we had been searching for, the signal we needed to predict the next input, was "location on the object." |
|
1 |
Hawkins - A Thousand Brains |
49 |
|
Different parts of the body (fingertips, palm, lips) might touch the coffee cup at the same time. The brain isn't making
one prediction:; it's making dozens or even hundreds of predictions at the
same time. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
The neocortex must know the location, relative to the cup, of every part of my body that is touching it. |
|
1 |
Hawkins - A Thousand Brains |
50 |
|
Vision is doing the same thing as touch. Each patch of your retina sees only a
small part of an entire
object. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
The brain doesn't process a picture, it starts
with a picture on the back
of the eye but then breaks
it up into hundreds
of pieces. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
The visual cortex assigns each piece to a location relative to the object being
observed. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
Since the complex circuitry in
every cortical column is similar, location and reference frames must be universal properties of the neocortex. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
Each column in the neocortex – whether it represents
visual input, tactile input, auditory input, language, or high-level thought
– must have neurons
that represent reference frames and locations. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
We need to think of the neocortex as primarily processing reference frames. Most of the
circuitry is there are to create reference frames and track locations. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
The brain builds models of the world by associating sensory input with locations and reference frames. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
A reference
frame allows the brain to learn the structure of something. |
|
0 |
Hawkins - A Thousand Brains |
50 |
|
A face is a nose, eyes, and mouth arranged in relative positions. You need a reference frame to specify the relative positions and structure of objects. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
By defining
an object using a reference
frame, the brain can manipulate the entire object at once. |
|
1 |
Hawkins - A Thousand Brains |
51 |
|
A car has many features in a range relative to each other. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
The brain only has to rotate or stretch the reference frame and all the
features of the
object rotate and stretch with it. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
A reference
frame is needed
to plan and create movements. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
Reference
frames also used in animated
films to run the characters as they move. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
The function of most of the neurons in each cortical column is to create reference frames and track
locations |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
Vernon
Mountcastle argued that there was a universal algorithm that exists in every cortical column, yet he
didn't know what the algorithm was. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
Francis
Crick wrote that we needed
a new framework to
understand the brain, yet he didn't
know what the framework
should be. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
Mountcastle's algorithm and Crick's framework were both based on reference
frames. |
|
0 |
Hawkins - A Thousand Brains |
51 |
|
Reference frames were the missing ingredient, the key to unraveling
the mystery of the neocortex and to understanding
intelligence. |
|
0 |
Hawkins - A Thousand Brains |
52 |
|
Insight that the neocortex is infused with the reference frames. |
|
1 |
Hawkins - A Thousand Brains |
53 |
|
The idea that the neocortex is infused with reference frames solve so many
constraints that I immediately knew it was
correct. |
|
1 |
Hawkins - A Thousand Brains |
53 |
|
Columns in the neocortex enable learning the structure of the world. |
|
0 |
Hawkins - A Thousand Brains |
53 |
|
A single cortical column could learn the
three-dimensional shape
of objects by sensing and moving and sensing and moving. |
|
0 |
Hawkins - A Thousand Brains |
53 |
|
We also showed how a column can recognize a previously learned object in the
same manner. |
|
0 |
Hawkins - A Thousand Brains |
53 |
|
We then showed how multiple columns in the neocortex work together to more quickly recognize objects. |
|
0 |
Hawkins - A Thousand Brains |
53 |
|
The entire neocortex works by creating reference frames, with many thousands active simultaneously. |
|
0 |
Hawkins - A Thousand Brains |
54 |
|
The idea that the neocortex represents locations and reference frames in every column was too exciting to
hold back. |
|
1 |
Hawkins - A Thousand Brains |
54 |
|
Neurons in the neocortex create reference frames by looking at an older part of the brain, the entorhinal cortex. |
|
0 |
Hawkins - A Thousand Brains |
54 |
|
Every cortical column in the neocortex creates reference frames. |
|
0 |
Hawkins - A Thousand Brains |
54 |
|
Neocortex learns a rich and detailed model of the world, which he uses to constantly predict what its next sensory inputs will be. |
|
0 |
Hawkins - A Thousand Brains |
54 |
|
Most predictions are represented by dendrite
spikes that temporarily
change the voltage inside a neuron and make a neuron fire a little bit sooner. |
|
0 |
Hawkins - A Thousand Brains |
54 |
|
Circuits in the neocortex that use the new neuron model can learn and predict sequences. |
|
0 |
Hawkins - A Thousand Brains |
55 |
|
A circuit can also predict the next sensory input when the inputs are changing due to all our movements. |
|
1 |
Hawkins - A Thousand Brains |
55 |
|
Each cortical column must know the location of its input relative to the object being sensed. |
|
0 |
Hawkins - A Thousand Brains |
55 |
|
A cortical
column requires a reference
frame that is fixed to the object. |
|
0 |
Hawkins - A Thousand Brains |
59 |
|
In mammals, the old brain parts
where the map creating neurons exist are the hippocampus and the entorhinal
cortex. |
|
4 |
Hawkins - A Thousand Brains |
59 |
|
In 1971, scientist John O Keith
placed a wire into a rat's brain. |
|
0 |
Hawkins - A Thousand Brains |
59 |
|
The wire recorded the spiking
activity of a single neuron and the hippocampus. |
|
0 |
Hawkins - A Thousand Brains |
60 |
|
The details of how to play sales
and grid cells work are complicated and still not completely understood. |
|
1 |
Hawkins - A Thousand Brains |
61 |
|
Humans, rats, indeed all
mammals, use the same mechanism for knowing our location. |
|
1 |
Hawkins - A Thousand Brains |
61 |
|
We all have grid cells in place
cells that create models of the place we have been. |
|
0 |
Hawkins - A Thousand Brains |
65 |
|
Every cortical column can learn models of complete objects. |
|
4 |
Hawkins - A Thousand Brains |
65 |
|
A cortical
column has multiple
layers of neurons. |
|
0 |
Hawkins - A Thousand Brains |
65 |
|
Altghough a cortical
column is tiny, about 1
mm on a side, each of
these layers might have 10,000
neurons. |
|
0 |
Hawkins - A Thousand Brains |
65 |
|
Upper layer receives the sensory input to the column. When it arrives, it causes several hundred neurons to become active. |
|
0 |
Hawkins - A Thousand Brains |
66 |
|
Bottom
layer represents the current
location in a reference
frame. |
|
1 |
Hawkins - A Thousand Brains |
68 |
|
Every cortical
column learns models
of objects. |
|
2 |
Hawkins - A Thousand Brains |
68 |
|
The columns do this using the same best basic
method that the old
brain uses to learn models
of environments. |
|
0 |
Hawkins - A Thousand Brains |
68 |
|
Each cortical
column has a set of cells the grid cells, another set equivalent to place cells, and another set equivalent to head direction cells, all of which were first discovered in parts of the old brain. |
|
0 |
Hawkins - A Thousand Brains |
68 |
|
Each cortical
column is small, about the width of a
piece of thin
spaghetti,
and the neocortex is large, about the size of a dinner
napkin. |
|
0 |
Hawkins - A Thousand Brains |
68 |
|
There are about 150,000 columns in the human cortex. |
|
0 |
Hawkins - A Thousand Brains |
69 |
|
Mountcastle proposed that every column in the neocortex performs the same basic function. |
|
1 |
Hawkins - A Thousand Brains |
69 |
|
Language, and other high-level cognitive
abilities, at some fundamental level are the same
as seeing,
touching, and hearing. |
|
0 |
Hawkins - A Thousand Brains |
69 |
|
Mountcastle deduced that there must be some
basic function that underlies everything the neocortex does – not just perception, but all the things we think of as intelligence. |
|
0 |
Hawkins - A Thousand Brains |
70 |
|
What kind of function, or algorithm, can create all
aspects of human
intelligence? |
|
1 |
Hawkins - A Thousand Brains |
70 |
|
Theory says that cortical columns create reference
frames for each
observed object. |
|
0 |
Hawkins - A Thousand Brains |
70 |
|
A reference
frame is like an invisible,
three-dimensional grid surrounding and attached to
something. |
|
0 |
Hawkins - A Thousand Brains |
70 |
|
The reference
frame allows a cortical
column to learn the
locations of features that define the shape of an object. |
|
0 |
Hawkins - A Thousand Brains |
70 |
|
We can think of reference frames as a way to organize any kind of
knowledge. |
|
0 |
Hawkins - A Thousand Brains |
70 |
|
Reference
frames can also be used to organize knowledge of things we can't directly sense. |
|
0 |
Hawkins - A Thousand Brains |
71 |
|
We have knowledge about concepts such as democracy, human rights,
and mathematics. |
|
1 |
Hawkins - A Thousand Brains |
71 |
|
All
knowledge is stored in reference frame. |
|
0 |
Hawkins - A Thousand Brains |
71 |
|
The brain arranges all knowledge using reference frames, and thinking is a form of moving. |
|
0 |
Hawkins - A Thousand Brains |
71 |
|
Thinking occurs when we activate successive locations and reference frames. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Reference
frames are present everywhere
in the neocortex. |
|
1 |
Hawkins - A Thousand Brains |
72 |
|
Every
column in the neocortex has cells that create reference frames. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Cells that create reference frames are similar, but not identical, to the grid cells and place cells found in older
parts of the brain. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Reference
frames are used to model
everything we know, not just physical objects. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
A column is a mechanism built of neurons
that blindly tries to discover and model the structure of whatever it is causing its inputs to change. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Brains first evolved reference frames to learn the structure of environments so they could move
about the world. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Then brain evolved to use the same mechanism to learn the structure
of physical objects so that they can recognize and manipulate them. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Brains once
again evolved to use
that same mechanism to learn and
represent
the structure underlying conceptual
objects,
such as mathematics and democracy. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
All
knowledge is stored
at locations
relative to reference
frames. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Reference
frames are not an
optional component of intelligence; they are the structure in which all information is stored in the
brain. |
|
0 |
Hawkins - A Thousand Brains |
72 |
|
Everything
you know is paired with the location and a reference frame. |
|
0 |
Hawkins - A Thousand Brains |
73 |
|
Organizing knowledge as reference frames makes the facts actionable. |
|
1 |
Hawkins - A Thousand Brains |
73 |
|
Thinking is a form of movement. |
|
0 |
Hawkins - A Thousand Brains |
73 |
|
If everything
we know is stored in reference frames, then to recall
stored knowledge we have to activate the appropriate locations in the appropriate reference frames. |
|
0 |
Hawkins - A Thousand Brains |
73 |
|
Thinking occurs when the neurons invoke location after location in a reference frame, bringing to mind what was stored in each location. |
|
0 |
Hawkins - A Thousand Brains |
73 |
|
The succession
of thoughts that we experience when thanking is analogous to the succession of sensations
we experience when touching an object with a finger. |
|
0 |
Hawkins - A Thousand Brains |
73 |
|
Reference frames in the Neo
cortex allow you to figure out the steps you should take to achieve more
conceptual goal. |
|
0 |
Hawkins - A Thousand Brains |
74 |
|
What and where pathways. |
|
1 |
Hawkins - A Thousand Brains |
74 |
|
The brain has two
parallel vision systems, called the what visual pathway and the where visual pathway. |
|
0 |
Hawkins - A Thousand Brains |
74 |
|
What pathway
is a set of cortical
regions
that starts at the very
back of the brain and moves around to the sides. |
|
0 |
Hawkins - A Thousand Brains |
74 |
|
Where
pathway is a set of
regions that starts out at the back of the brain but moves up toward the top. |
|
0 |
Hawkins - A Thousand Brains |
74 |
|
Similar parallel pathways also exist for other
senses. |
|
0 |
Hawkins - A Thousand Brains |
74 |
|
What and where pathways have complementary role. |
|
0 |
Hawkins - A Thousand Brains |
76 |
|
Reference
frames for concepts. |
|
2 |
Hawkins - A Thousand Brains |
76 |
|
Reference frames don't have to
be anchored in something physical. |
|
0 |
Hawkins - A Thousand Brains |
77 |
|
When a column learns a model of something, part of the
learning
is discovering
what is a good reference frame, including the number of dimensions. |
|
1 |
Hawkins - A Thousand Brains |
83 |
|
Language is arguably the most important
cognitive ability that distinguishes
humans from all of the animals. |
|
6 |
Hawkins - A Thousand Brains |
83 |
|
Without the ability to share knowledge and experiences via language, most of modern
society would not be
possible. |
|
0 |
Hawkins - A Thousand Brains |
84 |
|
Two modest
size regions of the neocortex are responsible for language. |
|
1 |
Hawkins - A Thousand Brains |
84 |
|
Wernicke's
area is thought to be responsible for language comprehension. |
|
0 |
Hawkins - A Thousand Brains |
84 |
|
Broka's
area is thought to be responsible for language production. |
|
0 |
Hawkins - A Thousand Brains |
84 |
|
We use spoken language, written language, and sign language. |
|
0 |
Hawkins - A Thousand Brains |
84 |
|
Wernick is at brokers areas don't get input directly from the sensors, so the comprehension of language must
rely on auditory and visual regions. |
|
0 |
Hawkins - A Thousand Brains |
84 |
|
The production
of language must rely on different motor abilities. |
|
0 |
Hawkins - A Thousand Brains |
84 |
|
Large areas
of the neocortex
are needed to create and understand language. |
|
0 |
Hawkins - A Thousand Brains |
95 |
|
Reference
frames and cortical
columns suggest a
different way of thinking about how the neocortex
works. |
|
11 |
Hawkins - A Thousand Brains |
96 |
|
All cortical
columns, even in
low-level sensory regions, are capable of learning and recognizing complete objects. |
|
1 |
Hawkins - A Thousand Brains |
96 |
|
A column that senses only a small part of
an object can learn a
model of the entire
object by integrating
its inputs over time. |
|
0 |
Hawkins - A Thousand Brains |
96 |
|
The neocortex has many models of any particular object. The models are in different columns. They are not identical, but complementary. |
|
0 |
Hawkins - A Thousand Brains |
96 |
|
A single
column can learn
hundreds of complex
objects. |
|
0 |
Hawkins - A Thousand Brains |
96 |
|
What a column
learns
is limited by its inputs. |
|
0 |
Hawkins - A Thousand Brains |
96 |
|
Even within a single sensory modality, such as vision, columns get different
types of input and will learn different types of models. |
|
0 |
Hawkins - A Thousand Brains |
96 |
|
There are some
vision columns that get color input and other vision columns that get black-and-white
input. |
|
0 |
Hawkins - A Thousand Brains |
96 |
|
A column in region V1 gets input from a very small area of the retina. |
|
0 |
Hawkins - A Thousand Brains |
97 |
|
Where is knowledge stored in the
brain? Knowledge in the brain is distributed. Knowledge of something is distributed in thousands of columns, but they are a small subset of all the columns. |
|
1 |
Hawkins - A Thousand Brains |
97 |
|
A column in
V2 gets input from a large
area of the retina, but the image
is fuzzier. |
|
0 |
Hawkins - A Thousand Brains |
98 |
|
A neuron never depends on a single synapse. Instead, it might use 30 synapses to recognize a pattern. |
|
1 |
Hawkins - A Thousand Brains |
98 |
|
A network
of neurons is never
dependent on a single
cell.
Even the loss
of 30% of the neurons usually has only a marginal effect on the performance of the network. |
|
0 |
Hawkins - A Thousand Brains |
98 |
|
The neocortex is not dependent on a single cortical column. The brain continues to function even if a stroke or trauma wipes out thousands of columns. |
|
0 |
Hawkins - A Thousand Brains |
99 |
|
The brain does not rely on one model of anything. Our knowledge of something is distributed among thousands of cortical columns. |
|
1 |
Hawkins - A Thousand Brains |
99 |
|
Each column is a complete sensorimotor system. |
|
0 |
Hawkins - A Thousand Brains |
99 |
|
How do our sensory
inputs get bound into
a singular percept? Instead of converging onto one location,
the connections go to every direction. We have proposed an answer: column
vote.
Your perception
is the consensus the columns reach by voting. |
|
0 |
Hawkins - A Thousand Brains |
102 |
|
Cells in some layers send axons long distances within the neocortex. We propose that the cells with long-distance connections are voting. |
|
3 |
Hawkins - A Thousand Brains |
102 |
|
The basic idea of how columns
can vote is not complicated. Using its long-range
connections,
a column broadcasts what it thinks it is observing. Often a column will be uncertain, in which case its neurons will
send multiple possibilities at the same
time. Simultaneously, the column receives projections from other columns representing their guesses. The most common guesses suppress the least common
ones until the entire network settles on
one answer. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
Recognizing an object means the columns now
agree on what object they are sensing. |
|
1 |
Hawkins - A Thousand Brains |
103 |
|
The voting
neurons in each
column form a stable
pattern that represents
the object and where it is relative to you. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
The activity of the voting neurons does not change as you move
your eyes and fingers, as long as they are sensing the same object. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
The other
neurons in each
column change with movement, but the voting
neurons, the ones that represent
the object,
do not. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
If you could look
down on the neocortex, you would see a
stable pattern of activity in one layer of cells. The stability would span large areas, covering thousands
of columns. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
These are the voting neurons. The activity of the
cells in other layers would be rapidly changing on a column by
column basis. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
What we perceive is based on the stable
voting neurons. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
The information from these neurons is spread broadly to other areas of the brain, where it can be turned into language or stored in short-term memory. |
|
0 |
Hawkins - A Thousand Brains |
103 |
|
We are not
consciously aware of the changing activity within each column, as it stays
within the column and is not accessible to other parts of the brain. |
|
0 |
Hawkins - A Thousand Brains |
104 |
|
The number
of voting neurons active at any time is small. |
|
1 |
Hawkins - A Thousand Brains |
104 |
|
If you were looking at the neurons responsible for voting, you might see 98% of the cells being silent and 2% continuously firing. |
|
0 |
Hawkins - A Thousand Brains |
104 |
|
The activity of the other cells in the cortical column
would be changing with the changing input. |
|
0 |
Hawkins - A Thousand Brains |
104 |
|
It would be easy to focus your attention on the changing neurons and miss the
significance of the voting neurons. |
|
0 |
Hawkins - A Thousand Brains |
104 |
|
The brain wants to reach a consensus. |
|
0 |
Hawkins - A Thousand Brains |
104 |
|
The voting
layer wants to reach
a consensus – it does
not permit two objects to be active
simultaneously – so it takes
one possibility
over the other. |
|
0 |
Hawkins - A Thousand Brains |
105 |
|
The brain
can attend to smaller or larger parts of the visual scene. Exactly how the brain does this is not well understood, but it involves
the thalamus, which is tightly connected to all areas
of the neocortex. |
|
1 |
Hawkins - A Thousand Brains |
105 |
|
Attention plays an essential role in how the brain learns models. |
|
0 |
Hawkins - A Thousand Brains |
105 |
|
What we
think is happening is that each time you attend to a different object, your brain
determines the objects
location
relative to
the previously attended object. |
|
0 |
Hawkins - A Thousand Brains |
105 |
|
As I look
around the dining room, my brain is not only recognizing all the
objects in the room but simultaneously determining where each
object is
relative to
the other objects in to the room. Just by glancing around, my brain builds a model of the room that includes all the objects that I attended to. |
|
0 |
Hawkins - A Thousand Brains |
105 |
|
Often, the models you learn are temporary. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
We are
constantly learning models of everything
we sense. |
|
1 |
Hawkins - A Thousand Brains |
106 |
|
If the arrangement
of features in our model stays fixed, then the model might be remembered
for a long time. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
If the arrangement changes, then the models
are temporary. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
The neocortex never stops learning
models. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
Every shift
of attention is adding
another item
to
a model of something. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
It is the same
learning process if the models are ephemeral or long-lasting. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
Hierarchy in the Thousand Brain Theory |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
The Thousand
Brain Theory says that a hierarchy of
neocautical regions is not
strictly necessary. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
Is the neocortex
organized a hierarchy or as
thousands of models voting to reach a consensus? |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
The anatomy
of the neocortex suggests that both hierarchy and thousands of models voting exist. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
We have proposed that complete objects, not features, are paseds between hierarchical
levels. |
|
0 |
Hawkins - A Thousand Brains |
106 |
|
The neocortex
uses hierarchy to assemble
objects into more
complex objects. |
|
0 |
Hawkins - A Thousand Brains |
107 |
|
The entire
world is
learned as a complex hierarchy of objects located relative
to other objects. |
|
1 |
Hawkins - A Thousand Brains |
107 |
|
We suspect that some amount of hierarchical learning occurs within
each column. |
|
0 |
Hawkins - A Thousand Brains |
107 |
|
Some
hierarchical learning will be handled by the hierarchical
connections
between regions. |
|
0 |
Hawkins - A Thousand Brains |
107 |
|
How much is being learned within the single column and how much is being learned in the connections between regions is not
understood. |
|
0 |
Hawkins - A Thousand Brains |
107 |
|
The answer will almost certainly
require a better understanding
of attention, which is why we are studying the thalamus. |
|
0 |
Hawkins - A Thousand Brains |
107 |
|
The thousand brain theory is
inherently a sensory motor theory. |
|
0 |
Hawkins - A Thousand Brains |
107 |
|
The relatively
large size of regions V1 and V2 in primates and the singularly large size of the region V1 in mice makes sense in the Thousand Brain
Theory
because every column can recognize
complete objects. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
Thousand
Brain Theory says that most of what we think of
his vision occurs in regions V1 and V2. The primary and secondary touch-related regions are also relatively large. |
|
1 |
Hawkins - A Thousand Brains |
108 |
|
Thousand
BrainTtheory can explain how neurons know what their next input will be while the eyes
are still in motion. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
Each column has models of complete objects and therefore knows what should be sensed on each
location on an object. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
If a column knows the current location of its input and how the eyes are moving, then it can predict the new location and what it
will sense there. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
Thousand
Brains Theory says there are thousands of models of every object. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
The pattern projected to the region V1 can be distorted and mixed up and it won't matter, because no
part of the neocortex tries to
reassemble the scrambled
representation. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
The voting
mechanism of the Thousand
Brains Theory explains why we have a singular non-distorted perception. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
Thousand
Brains Theory also explains how recognizing an object in one sensory modality leads to predictions in other sensory modalities. |
|
0 |
Hawkins - A Thousand Brains |
108 |
|
Thousand
Brains Theory shows how the neocortex can learn three-dimensional
models of objects using reference frames. |
|
0 |
Hawkins - A Thousand Brains |
109 |
|
The details of how place cells and grid cells create reference frames, learn models of environments, and plan behaviors are only partially
understood. |
|
1 |
Hawkins - A Thousand Brains |
109 |
|
This is an area of active research for both experimental neuroscientists and theorist. |
|
0 |
Hawkins - A Thousand Brains |
142 |
|
Michael
Graziano, Consciousness and the Social Brain, Princeton neuroscientist. |
|
|
|
|
|
|
|
|