Quantifying consciousness

Posted comment on ´The One-Second Test of Consciousness` by A. Ananthaswamy and published in New Scientist 20th February 2016 issue no. 3061 p. 10

SUMMARY

Ananthaswamy describes in his article how scientific groups are beginning to formulate a method by which the conscious experience can be quantified. It involves calculating the value of phi which is defined as the level of integration of information of an informational processing system. Therefore, the hypothesis can apply to consciousness which follows the Integrated Information Theory. This theory, according to Tononi and described in Ananthaswamy`s article, says that each aspect of what we are aware of is unified and hence, it is difficult to be aware of a single aspect of an experience since the brain integrates the sensory data into one. For the system to be conscious according to Tononi the integrated information must be greater than the sum of the individual elements. Phi is, therefore a measure of the success of this integration.

Ananthaswamy reports in this article on one approach to calculating phi which involves how each part is dependent on the other. The ´cruellest` division designated by phi is given the value of zero and is attributed to non-conscious status. This is where the parts are the least dependent on each other. As dependency increases so does the value of phi. Although the concept is simple, the execution of it is very difficult, but Tegmark, an American cosmologist, however has devised a fast way of approximating it.

Tegmark in his method regards each neuron as a node and the interconnections between it and other cells as links. Each link is given a thickness proportional to the strength of the interconnection. The thinnest links in the network are then disregarded and this action is repeated resulting in a step by step occlusion method until a single interconnected network remains. Cutting this interconnection into two would approximate to the ´cruellest cut`.  By carrying out this method the time taken to find phi is drastically shortened eg. finding out phi for the human brain takes less than a second and hence, the technique can be applied to reveal the level of human awareness at any one time. Gazzaniga has already seen an application for the method in research and therapy of neuropsychiatric disorders. He suggests that by measuring phi in each brain hemisphere he can determine whether each side retains its own consciousness unaware of the other and expects that if this is the case then the hemispheric values of phi will be lower than for the control unseparated brain.

Following verification of the hypothesis, the method could also be used to identify people with consciousness disorders who may have been misdiagnosed. Ananthaswamy gives the example of the misdiagnosis of minimally conscious state for someone who is conscious, but completely paralysed. In this case, high values of phi should be obtained. Others in the neuroscience field are also enthusiastic about the new method for the quantification of phi and consciousness.

COMMENT

Ananthaswamy`s article describes one method of putting a mathematical value to the conscious experience and hence, quantifying it. The method described is that of Tegmark who based it on Tononi`s Integrated Information Theory and it consists of the mathematical value of phi depicting neural activity of one event using multiple characters eg. colour and thickness of lines representing the neuronal firing. Therefore, from a neuroscientific perspective it is probably easy to find the zero value of phi which represents unconscious and this creates a ´symbolic` lowest level value which can be expanded when there is conscious awareness of events. However, problems can be predicted with this method of identification of the conscious experience in the case of awareness since the method requires the disregarding of the weakest connections of the neural image obtained.

The first problem is that everyone is individual and therefore, controls of neuronal experience and awareness would be required. It would have to be known what firing patterns are observed for an individual under normal conditions in order to determine if changes are observed. Would the sight of a red, round ball in the left peripheral field give a distinct neural network firing pattern which would be drastically different to one if the ball is observed straight on? It is likely that for any event, huge neuronal areas would be alight, representing interconnected systems, event characteristics, emotional values and memories. The neuronal firing representing the characteristics of which we are aware of consist of only a small part of this. Conscious awareness means those characteristics that a person can report, but other areas systems also form part of the neural firing signature observed at that time. This view is supported by the consciousness theories such as Zeki microdomains of consciousness, 3d default space of Jerath as well as Local Recurrence Theory and Reentrant Dynamic Core Theory. Therefore, a quantitative assessment of phi would include not only the characteristics of which we are consciously aware, but a mass of unattended system-related firing that allows that conscious awareness to take place. The problem with this method of phi identification is that the firing patterns of these essential, but unconscious systems, may be stronger than the one depicting the red colour or rounded shape if we consider our example of the red, round ball conscious event.  In this case, by disregarding the weakest firing the question is are we going to end up with only the attentional engagement mechanism for example and not the features of the conscious event?  This may be acceptable since firing relating to employed systems is likely to be common in multiple individuals and could theoretically be excluded from the original firing images before the process of removing weakest connections begins. However, the level of attention may vary individual to individual and therefore, firing levels of even this system may also differ making it difficult to set up a computer or mathematical model to disregard these weakest links.

Another problem of the method that has to be considered is the delay between event presentation and conscious awareness. The event characteristics are observed 200-300millisecs before conscious awareness (the Libet delay). Therefore, when should the firing image for the calculation of phi be taken? If it is taken at a time point 200millisecs or later after the presentation of the image as would be indicative for conscious awareness, the firing experience may already be non-representative of the actual experience. Some firing neurons are already going into their refractory periods leading to firing of other neurons according to priority to the unattended, lateral inhibition visual rules for example and even reporting the conscious event will strengthen some firing and not others. Therefore, firing patterns change with time and this correlates to parallel drafts being edited as part of Dennett`s  Multiple Drafts Theory and the victorious assembly with groups constantly competing and modulating according to Edelman and Tononi`s Reentrant Dynamic Core theory. Firing patterns also change from top-down influences. Both from a consciousness perspective (Ramachandran`s filling in) and from the neuroscientific perspective of the influence of memory recall, the neural signature is affected by the application of information of reactivated memories to the conscious experience. Hence, phi quantification is affected not only by bottom-up input, but also top-down and therefore, is that neural pattern being observed and ready to serve the mathematical quantification of consciousness representing just the red, round ball or the red, round ball loved as a child?

Therefore, we have seen that the quantification of phi is difficult to measure from a neuronal firing perspective. It may be possible if we could take several images of different circumstances that vary in only one simple, single sense characteristic eg. red colour or silhouette shape and compare these  removing consistent firing areas independent of actual firing level. If we assume that those firing links removed represent the common cognitive systems employed by the individual, the firing remaining should represent the single characteristic of which the individual is conscious. This image could be tested against the presentation of the same characteristic at such a speed that it remains unconscious, or is presented in the peripheral view or if possible by using a participant who suffers from blindsight. The problem comes when the method is applied to multiple characteristics for one single sensory event and is even greater if multiple senses are employed. Mercel and slippage will also present a problem with integrated firing in a multisensory temporal binding window, phase locked according to Walter Freeman`s dynamic systems approach, but with a time delay for some sensory information and not for others. And who is to say that this is the same for everyone in every single circumstance?

Therefore, Tegmark`s method for the quantification of phi looks possible and advantageous on paper, but a couple of examples from the neuroscientific perspective shows that actual physical imaging and then interpretation of those images present problems. Therefore, is phi a psychological concept only? From the neuroscience view, there is no denying that Integrated Information Theory applies to neuronal firing and integrated firing leads to the formation of assemblies and perception for example, but the constant modulation of those assemblies due to even physiological restraints makes it difficult to use the information in a generic situation. Simplicity of the conscious event may aid the neural representation imaged, but conscious awareness is a fleeting event, delayed in real-time, influenced by top-down as well as bottom-up systems and only possible if proven by language or action which changes the overall picture observed. All of these factors will lead to changes in the neural pattern that is observed. Therefore, it may be that phi can only be defined at the zero unconscious level or default network level in humans in which case, it must be asked what value does an accurate calculation of phi actually have apart from saying something is not conscious?

Since we`re talking about the topic…………………………………………..

…Transcranial magnetic stimulation (TCMS) is said to distinguish between patients who are described as minimally conscious to those suffering from unresponsive wakefulness syndrome presumably by the differing responsiveness of the auditory, olfactory and touch senses and deep TCMS is said to decrease levels of self-awareness. Therefore, could TCMS be used to create the control images for the calculation of phi?

….if 40HZ (gamma) oscillations are disrupted accurately in distinct brain areas could differences in neural patterns because of lack of synchronicity and hence phi be measurable?

….can we assume that seeing external events and actions performed by ourselves reflected in a mirror will produce different neural patterns to that observed directly and could these be used to ascertain the value and differences in phi?

 

Advertisements
This entry was posted in consciousness, Uncategorized and tagged . Bookmark the permalink.