Kurzweil Response

“The neocortex is responsible for sensory perception, recognition of everything from visual objects to abstract concepts, controlling movement, reasoning from spatial orientation to rational thought, and language…” Kurzweil works with this notion of hierarchal patterns (written, spoken, and visual) when explaining the ways of thinking.  He is shaping one’s visual and perceptual understanding of brain activity, and essentially creating a form for thought – “…the goal of the effort is to refine our model to account for how the brain processes information to produce cognitive meaning.” Therefore, it is important to read the text with an understanding of how one must abstract reality in order to explain it.

This idea goes hand in hand with how he relates to pattern recognition in memory and language. He notes that our memories are organized as patterns of lists. These patterns are generated through experience and given a certain weight due to individual values. However, these memories are not necessarily visual.  When recalling a memory and relaying it to another, these patterns aggregate and reconstruct a memory though a visual or linguistic imagination –  “you will essentially be reconstructing…images in your mind, because the actual images do not exist.”

This acquisition of knowledge creates a fundamental understanding of how the neocortex operates through one’s reality. Its potential to shape one’s understanding of the self parallels the potential to alter one’s reality. Over time, we will see how the evolution of understanding the brain may change our associations with each other and with ourselves.  It’s the ongoing conversation between scientists and technologists that is most beneficial in understanding our brains. It is this dialogue that will keep us from overly formalizing the brain, as it will open up a space for the abstract thinking.

Final Proposal – Ayo + Julie

Precedence:

The Turing Normalizing Machine

http://mushon.com/tnm/

Proposal:

To create a new classifier such as “surprise” we would need to get a baseline of raw EEG data. We will then introduce a the subject to a surprising stimulus and measure the difference in raw EEG data. We would need to repeat the process several times and have an average of the surprise stimulus data to be confident in the results. The problem here would be desensitisation to the stimulus. Over time the subject will be aware of the inducement and may fail to provide the appropriate surprise data. To counter this issue of desensitisation we would use multiple subjects. In addition to solving the desensitisation issue, we generate a more universal notion of the emotional state / classifier we are hoping to achieve. By analysing the patterns emerging from multiple surprise data response from different people, we can start to create a module for understanding a normalized physiological response based on a particular stimulus.

Brainstorm of ideas:

-       baseline of surprise

-       use a sweat sensor and the thinkgear

-       raw eeg?

-       surprise by playing a video

-       measure heart rate

-       emotional state that can be repeatedly tested

-       emotions to consider: anger, fear, nervous, agitation(annoyance), sad, apathy, empathy, disgust

-       have ppl look at optical illusions as stimulus?

-       trigger: disgust, primal instinct, visceral, elicit instinct, fear

-       map body’s physical chemical reaction with brainwave data

-       use a person without stimulus as control

-      test more ppl with stimulus

-       possibly use brainwaves of other beings besides humans

-       And It gives a universal response

-       Stimulus, sound and video

-       creating a new classifier, problem of being desensitized

-       take difference btwn all the testers and come up with an average

Brainesthetics

Our goal is to create something that is aesthetically valuable using brainwaves. How can you influence the form, color and motion of piece of art or a new media work? How can you personalize a piece by mapping your attention levels? How can we embody this idea of an immersive experience in the form of a series of experiences and iterations using brain data?

We also came across the idea of Neuroesthetics in our research, which is an attempt to combine neurological research with aesthetics by investigating the experience of beauty and appreciation of art on the level of brain functions and mental states.

And in reference to our first iteration, we also put forward the idea of creating a piece of art from scratch by programming a drawing tool that functions with neuro data.

9781780673479.in07 (1)

So we are currently exploring the making of a series of three pieces that would reflect these notions. Our precedents and inspiration stem from visually beautiful or intriguing media.

mountainLake_seamSort

 

 

Final Proposal – Maxine & Stephanie

We are interested in producing a physical output/interpretation of the brain waves through a wearable piece. Something along the same lines as the GER Mood Sweater – http://sensoree.com/artifacts/ger-mood-sweater/ . This item basically categorizes different GSR levels and assigns a color to correspond with it. Another similar project is the intimacy dress – http://www.studioroosegaarde.net/project/intimacy-2-0/.  This dress becomes transparent based on the heartbeat of the wearer. Is it possible to communicate with someone solely through a ‘brain language’ – much like ‘body language’?

5634286849_dd3491dfc4_z

What if you could display different levels of meditation through a similar kind of color indicator system? Would this system be useful ? We are also interested in how this type of wearable could affect your brain waves. Would your brain behavior change based on the sole fact that you are wearing this device? We are interested in exploring how these factors all feed into each other and what the implications of that are.

Chapter 3 Reading Response – Julie Huynh

How to Create a Mind: Chapter 3 Response

I think Kurzweil’s thesis poses a strong theory on how the brain processes pattern recognition.  Memories such as deja-vu fit into this theory where although you cannot specifically place the direct connection, the brain confabulates a connection triggered by undirected thoughts.  The neocortex confabulates a connection caused by the trigger of undirected thoughts, such as a smell could trigger a familiar memory from the past.  Dreams are examples of undirected thoughts that are linked with direct thoughts, which are a sequence of patterns.  However, our dreams, lucid dreaming fit outside of this theory because our brain is in a state where it can be lax on the cultural rules enforced by the neocortex, and it can explore consequences of actions without the repercussions and triggering the fear from the amygdala.  Interpretation of the pattern input of dreams would give us a peak into understanding unresolved fears and desires, but I think a different system would be necessary to understanding dream patterns.  The PRTM thesis relates to artificial intelligence because by emulating these pattern recognizers and the system that house them we could create artificial intelligence that could recognize speech, or in the future process information for us.  By creating programs to recognize phonemes, recognizing the spoken word using hierarchical hidden Markov models will incorporate the distribution of magnitudes of each input, on the speech level, this will computationally simulate human intelligence.  Starting with the simulation of speech recognition we can take the next step towards artificial intelligence with a cortical algorithm in the temporal of constituent lists.

Thoughts on creating a ‘mind’

According to Kurzweil’s  theory, the brain learns and recognizes patterns, and then arranges this information in hierarchies that make it possible to recall ‘memories’ and make associations in order to understand how language works. But memories have a certain fuzziness to them. What is truth? What are lies? Do these things make up your identity? Somewhere between what you experience in the world and how you act in it, is a space for processing information and building knowledge. Memories are our interpretations of our experiences versus direct representations of our actions. They seem to be represented in the brain as networks of related concepts. One hypothesis is that a healthy memory system copes with massive amounts of information by forming connections between concepts based on associations we’ve developed through experience. (http://musicandmemory.org/) Then, when we’re trying to remember a piece of information our brain  pulls up an assortment of associated concepts which ‘feel’ right, and thus, our cognitive experience is solely embedded in our understanding of the world as related to the stories we create for it. When trying to understand the origin of memories, scientists are led to this topic of synapses and their electrochemical relationships. Thus, memory, and therefore , language is regarded as means of expressing ones identity. There are many languages besides those that are written or spoken. By learning a new language, a person acquires a new way of knowing reality and of passing that knowledge onto others.

It’s curious to me that we automatically associate artificial intelligence with replicating the human brain. Isn’t there validity in modeling a programming system after a different kind of intelligence? Maybe one that is a hybrid of complex systems of thought that include other species. Thinking in this post-human way, we can start to speculate the future of memory enhancement capabilities that go beyond genetic modification/engineering. Only then can the future begin to enter the transhumanist world of cybernetics. Many people (ie. Ray Kurzweil) believe the absorption of technological devices into our bodies, to improve our day-to-day lives, is an inevitable next step. How would altering ourselves in this way change our identity? And furthermore, how would it effect our connection to other humans?

Notes on Kurzweil

Communications between cells happens purely on electrical and chemical basis.

The neocortex is present only in the mammalian species. The structure is hierarchical and is responsible for sensory perception, and in humans, rational thought. It is wrinkled to provide the most surface area, it constitutes the majority of our frontal lobe. The fundamental structures of the neocortex are remarkable uniform, it consists of singular mechanism repeated over and over again in a physical space of 2.5mm. The complexity of the system comes about by the intricate way it folds in on itself and the elaborate connectivity of the neural axons and dendrites. In order to understand the complexity of the system we need to develop abstract models of representation. In so doing we are creating and interface to manage and comprehend the complexity that arises from the 300+ pattern recognizers present in the neocortex.

Our patent recognizers work in parallel, although not firing all at once. Compared to computers, the pattern recognizers in our neocortex are slower. Where computers are faster at serial processing, their ability to recognize patterns is extremely limited. On the contrary, we are extremely adept at pattern recognition, but serial & logical processing must be learnt and deduced from patterns. We use the redundancy of recognizers to form comparisons between our numerous recognized patterns in order to create logic and think rationally.

In respect to the direction of data flow in the hierarchy of the neocortex, it is important to note that information can travel both up and down the hierarchy. Based on thresholds the neuron may fire, these thresholds may change based on weather or not the particular neuron is primed for it’s particular pattern. This threshold fungibility allows for the prediction of patterns, but may also allow for the mistaken recognition of patterns in cases where a prediction is proven false.

Where communication in the neocortex travels in a 1 dimensional array that utilizes lists for organization, the information (stimulus) may reach the brain from a 2,3 and at time 4 dimensional sensory input. With respect to memory, they have no “named” patterns, but must be triggered by other patterns, subsequently, additional memories and patterns are triggered… in essence recreating the moment of that particular memory. Like and orphaned web page, if no patterns or memories connect to a particular memory, it cannot be retrieved. In this process the notion of redundancy is important, as alternate memories or recognized patterns may be used to retrieve a seemingly unretrievable memory, even though the memory of the event may not be recollected.

The hierarchy of language simulates the hierarchy of the brain, which in turn simulates the hierarchy of nature.

Love this reading… and could write more… but think I will stop here.

I like Kurzweils theory on PRTM and think his thesis is valid, I believe there are missing element… but feel his description of a great foundation for the lay person. Getting deeper into the particular biological, physical, electrical and neurochemical processes would be great… but think this might be better suited for a neuroscience student. As he describes about the need for abstracted models to understand complexity… in the same manner, his abstracted method of description allows for access to the complexity of a system such as the brain.

The systems that Kurzweil describes in this chapter is quite similar to parallel computing. But the majority of computer processing is done in series. By having computers perform pattern recognition one may be able to emulate the systems and performance of the brain thus giving rise to better speech recognition, computer vision or, the holy grail of artificial intelligent machines. In my opinion, given the exponential growth in computation, i believe this achievement is accessible in the not far future.

Ray Kurzweil Reading

The book, How to Create a Mind: The secret of Human Thought Revealed, chapter 3: A Model of The Neocortex: The Pattern Recognition Theory of Mind, the author, Ray Kurzweil wrote about how the complicated and intricately brain works. Some points were very interested to me and raised my attentions.

Kurzweil wrote the lower levels of the neocortex is the redundant patterns, and the higher levels is patterns with all sorts of continuums, such as levels of attractive, irony, happiness, frustration, and myriad others. Also, the methods of the patterns recognizer are 1) receiving sensory data 2) redundancy in our cortical pattern memory 3) ability to combine two lists. He wrote creating and gaining connections of these three methods are coming from experience.

I agree with Kurzweil. We as individuals, we have all different memories and that makes us unique. We do not experience smelling, hearing, feeling, thinking, perceiving, and predicting something same ways. As human being, these unique experience are the major difference from being AI.

Also, I found that we start to have experience and learning from when we are the fetus. I have heard many time that how mother thinkings, feelings, and acting effect on babies. I agree that the fetus has a brain and they learn and recognize directly, and that develop their memory stronger.

“Dream are examples of undirected thoughts” (p.143) I have heard that you are the most creatively brainstorm while you are in sleep because you are blocked from real world matters which is unnecessarily regulate your thoughts.

Response to How to Create a Mind – Betty

Ray Kurtzweil’s Pattern Recognition Theory of the Mind proposes that rather than using deductive reasoning, humans use pattern recognizers to create connections between neurons that allow them to learn. Chapter 3 of How to Create a Mind concentrates on the neocortex, which Kurtzweil argues is made up of cortical columns that contain a total of 300 million pattern recognizers. According to Kurzweil, our neocortex is a blank slate when we are born and only our experiences can wire connections between pattern recognizers. His hypothesis states that the structure of the neocortex is malleable, changing the hierarchical way it is connected between modules as we learn over time.

I particularly thought that this argument was convincing in terms of the examples he gave:

[Translating memories into language] is also accomplished by the neocortex, using pattern recognizers trained with patterns that we have learned for the purpose of using language. Language is itself highly hierarchical and evolved to take advantage of the hierarchical nature of the neocortex, which in turn reflects the hierarchical nature of reality.”

If reality truly is hierarchical, as Kurtzweil argues, then it would be natural that the way our brain processes information would be hierarchical as well.

This chapter could easily have been extremely technical — however, overall, he explained the concepts of hierarchical pattern recognition clearly, with easily understood examples, such as a higher/lower level recognizers sending a message of how a loved one is recognized.

Kurtzweil’s pattern recognition theory of the mind draws many parallels between the way we process and learn things and the way he believes computers (artificial intelligence) learn. Ultimately, Kurtzweil makes a case that one day we will be able to imitate and perhaps venture further than human intelligence. While I personally do not believe that the brain’s structure is just a simple algorithmic process as he implies, it did start me thinking of the capabilities of artificial intelligence, and if we did create them, how far should we take it?

For example, here is one quote that made me begin to think of how “human” our potential AI’s should be:

“However, if I don’t think about her for a given period of time, then these pattern recognizers will become reassigned to other patterns. That is why memories grow dimmer with time: The amount of redundancy becomes reduced until certain memories become extinct.”

If “memories become extinct”, how can an AI imitate that? Should it imitate that? With all the distortions in our logical thinking, how much of human intelligence would we want to imitate?

Imagine if we created artificial intelligence that is capable of being exposed to as many experiences and sensory inputs that we are bombarded by everyday, but has a pattern recognizer less prone distortions of memory that are natural in humans…I was hoping that Kurtzweil touch on the moral responsibilities that we have in creating artificial intelligence, but unfortunately, he did not dive into the potential dangers of artificial intelligence.

Can you truly replicate human intelligence? You can replicate the logical process, perhaps, but can you replicate empathy? Emotion? While I don’t doubt that one-day AIs will have the ability to surpass our own intelligence, I doubt that we can replicate how the human mind truly works.

POV on Kurzweil’s Pattern Recognition Theory of Mind

This is a review on Chapter 3, in Ray Kurzweil’s “How To Create A Mind”: The task of reverse engineering the brain might be one of the biggest tasks in human civilization, and is one that Kurzweil casually approaches in his new book “How to Create a Mind: The Secrets of Human Thought Revealed”. Kurzweil’s work has undoubtedly been instrumental to the field of AI, and as a pioneer in speech recognition technologies he is the right man to consult on the topic.

That the brain understands patterns through a hierarchal grid system is nothing new, and humans have been under this assumption since the late 1950’s. We have been able to create pattern recognizers able to perform intelligent tasks to a certain extent, IBM’s Watson clearly being the best performer (although it’s parameters where very closed and focused) and Google only last year tried to create an algorithm using the same logic, but only reached an accuracy rate of about 18%. Yann LeCun, the N.Y.U. researcher who was just appointed to run Facebook’s new A.I. lab mentioned the that: “AI [has] ‘died’ about four times in five decades because of hype: people made wild claims (often to impress potential investors or funding agencies) and could not deliver. Backlash ensued. It happened twice with neural nets already: once in the late 60’s and again in the mid-90’s.” The point is, that the brain learns by chunks of low level to high level patterns is not the interesting part – the interesting and still unknown part is how?

That we are able to replicate human thought with AI using this logic is yet to be proven, which leads to the main criticism I have – why did he not test his theory like any good scientist would do? He seems to make a lot of claims about how this theory would solve everything with very little evidence. Moreover, he claims to reveal the secret of the human brain, thought, and emotion without fully making the connections. He is clearly merely interested in simply replicating certain “thought processes” mathematically.

Much of his logic I would agree to, for example that a computer is much more able to process the 300 million pattern recognizers a human brain presumably has (how he came to this number is unclear). Also the idea that there are directed and undirected thought processes, which explain our state of dreaming or mediation vs. goal oriented thinking makes much sense and pertains to the variables we are currently working with in this class – attention/meditation with various BCI technologies are to date the most reliable. The way in which information flows down the conceptual hierarchy as well as up thus enabling us to predict the future is also very plausible,  however how we are able to do this or remember new patterns is unclear.

Over all, the PRTM indeed gives the unfamiliar some thoughts on how we would begin computing a human brain. However nothing new is revealed or tested, as he even falls short in proving the differences between his and Hawkins theory. Actually, many of his unsupported ideas may in fact merely be … dreamt up.