Pragmatic Information
In order to derive meaningful information related to the output of EEG signals based on the input of diverse kind of stimuli, we require a methodology capable of processing complex signals that are highly dynamic and non linear in nature. This kind of information in the brain, the one that carries meaning and knowledge for action and decision making is content dependent and has no semantic markers attached to it. When analysing EEG or ECoG signals we seek to derive their content for action instead of the semantic integrity of the message which is non-existent and from which we are unable to derive any information content associated to such a semantic marker.
The work of Shannon and Weaver provided us with the means to measure the information transferred through a communication channel and therefore guaranty to an extent that the semantic integrity of the message was preserved from sender to receiver. However, this kind of measure, the Shannon index, has no means to provide us with the information content of the message and the meaning it carries for action. This makes it necessary to use other kinds of indexes more suited for the task at hand, hence the use of pragmatic information indexes.
From semiotics we learn that pragmatic information exists in association with intention, decision-making and action, like for example, when we say something that is meant to produce a certain interaction with others and the environment. This is different than semantic information, which is only associated with the grammar and linguistic ingredients when we say something. Pragmatic information can be found in many forms like: (1) the information content (meaning for action) associated to the spoken word, (2) body language even when nothing is said, (3) biological markers like stress hormones or amplitude wave patterns as in brain waves or laser wave-like patterns. The bottom line distinctions to learn is that pragmatic information is created together with the creation of meaning and intention before we initiate an action such as speaking or pointing our finger to a place we have identified as relevant to us in a certain way, for example, the school where I am picking up my beloved child who is important to me. Semantic information on the other hand is just associated to the grammatical and linguistic code that comprises the message.
The work presented here is inspired by and related to previous studies on nonlinear spatio-temporal dynamics of rabbit brain data conducted at W.J. Freeman’s Lab at UC Berkeley. The concepts and methodology presented here has its origin and foundation in the work of Walter Freeman together with Robert Kozma and more recently with the collaboration of Jeffery Jonathan (Joshua) Davis and the team at the Embassy of Peace, Whitianga, New Zealand.
At the present time we continue to explore more ways of improving such a methodology. These efforts aim toward establishing a comprehensive approach to be applied to brain dynamics in a wide range of practical fields, to enhance human capabilities, to provide support for the disabled in order to improve well-being and peaceful living.
The work of Shannon and Weaver provided us with the means to measure the information transferred through a communication channel and therefore guaranty to an extent that the semantic integrity of the message was preserved from sender to receiver. However, this kind of measure, the Shannon index, has no means to provide us with the information content of the message and the meaning it carries for action. This makes it necessary to use other kinds of indexes more suited for the task at hand, hence the use of pragmatic information indexes.
From semiotics we learn that pragmatic information exists in association with intention, decision-making and action, like for example, when we say something that is meant to produce a certain interaction with others and the environment. This is different than semantic information, which is only associated with the grammar and linguistic ingredients when we say something. Pragmatic information can be found in many forms like: (1) the information content (meaning for action) associated to the spoken word, (2) body language even when nothing is said, (3) biological markers like stress hormones or amplitude wave patterns as in brain waves or laser wave-like patterns. The bottom line distinctions to learn is that pragmatic information is created together with the creation of meaning and intention before we initiate an action such as speaking or pointing our finger to a place we have identified as relevant to us in a certain way, for example, the school where I am picking up my beloved child who is important to me. Semantic information on the other hand is just associated to the grammatical and linguistic code that comprises the message.
The work presented here is inspired by and related to previous studies on nonlinear spatio-temporal dynamics of rabbit brain data conducted at W.J. Freeman’s Lab at UC Berkeley. The concepts and methodology presented here has its origin and foundation in the work of Walter Freeman together with Robert Kozma and more recently with the collaboration of Jeffery Jonathan (Joshua) Davis and the team at the Embassy of Peace, Whitianga, New Zealand.
At the present time we continue to explore more ways of improving such a methodology. These efforts aim toward establishing a comprehensive approach to be applied to brain dynamics in a wide range of practical fields, to enhance human capabilities, to provide support for the disabled in order to improve well-being and peaceful living.
In recent years the use of pragmatic information has extended to many fields. One of them is the field of physics, particularly in the study of "Pragmatic information and dynamical instabilities in a multimode continuous-wave dye laser" as studied by H. Atmanspacher, H. Scheingraber, in order to show that the concept of pragmatic information is relevant in describing self-organizing systems via two approaches such as synergistic and non-equilibrium thermodynamics. They have also pointed out the relevance of pragmatic information in relation to meaning and complexity.
In terms of brain dynamics, the concept of pragmatic information needs to be translated into a mathematical form or structure (in our case a ratio) that can be representative of the meaning and the knowledge encoded in brain signals. Usually, brain signals are measured via EEG or ECoG and though differently, they carry information in multiple frequency bands, about the activity of the brain generated by a large set of stimuli both controlled (like in the lab) as well as uncontrolled (environmental and unconscious or instinctive). These signals are usually measured on multiple locations with electrodes (channels) and such measurements may cover either small or large areas of the brain, depending on research or clinical interest. The spacing between electrodes may be very small or large (spatial resolution) which together with the evolution of the signal in time will produce an image and data suited for spatio-temporal analysis. In order to do that for each band, the original signal has to be band pass filtered for a particular narrow or broad range of frequencies. After filtering, the signal is Hilbert transformed and that allows for the generation of a set of new signals as follow: analytic amplitude (AA), analytic phase (AP) and instantaneous frequency (IF) based on certain computations.
Let's say we measured 64 channels on the scalp of a human being in the region of the pre-frontal and frontal cortex with a dry electrodes array with 5 mm spacing between each probe and represent the signal as the matrix S(t). The EEG signal Sj(t) of each channel (j=1,…,64) is band pass filtered and a new signal Yj(t) is obtained. This signal is then transformed via the Hilbert transform to a time series (signal) of complex numbers, Zj(t) (2.1), with a real part, vj (t) equal to Yj(t) and an imaginary part, uj (t). After the Hilbert transform of Y(t) is calculated, the modulus (2.2), the angle (2.3) and IF (2.4) are computed as follows:
In terms of brain dynamics, the concept of pragmatic information needs to be translated into a mathematical form or structure (in our case a ratio) that can be representative of the meaning and the knowledge encoded in brain signals. Usually, brain signals are measured via EEG or ECoG and though differently, they carry information in multiple frequency bands, about the activity of the brain generated by a large set of stimuli both controlled (like in the lab) as well as uncontrolled (environmental and unconscious or instinctive). These signals are usually measured on multiple locations with electrodes (channels) and such measurements may cover either small or large areas of the brain, depending on research or clinical interest. The spacing between electrodes may be very small or large (spatial resolution) which together with the evolution of the signal in time will produce an image and data suited for spatio-temporal analysis. In order to do that for each band, the original signal has to be band pass filtered for a particular narrow or broad range of frequencies. After filtering, the signal is Hilbert transformed and that allows for the generation of a set of new signals as follow: analytic amplitude (AA), analytic phase (AP) and instantaneous frequency (IF) based on certain computations.
Let's say we measured 64 channels on the scalp of a human being in the region of the pre-frontal and frontal cortex with a dry electrodes array with 5 mm spacing between each probe and represent the signal as the matrix S(t). The EEG signal Sj(t) of each channel (j=1,…,64) is band pass filtered and a new signal Yj(t) is obtained. This signal is then transformed via the Hilbert transform to a time series (signal) of complex numbers, Zj(t) (2.1), with a real part, vj (t) equal to Yj(t) and an imaginary part, uj (t). After the Hilbert transform of Y(t) is calculated, the modulus (2.2), the angle (2.3) and IF (2.4) are computed as follows:
From these computations we get the basic type of signals for each channel that will be used in the next step of the analysis. These signal are, as mentioned before AA, AP and IF which are equal to Aj(t), Pj(t) and Fj(t) respectively. With these signals we compute a set of indices (RSA, RAA, SAA) for sub-windows of time, which will be also used in the computations of the pragmatic information indexes. All of these signals and indices can be plotted for visual analysis.
The aim of this methodology is to allow us to perform a thorough analysis based on multiple runs and rabbits in order to get statistically significant results, as more data becomes available, in order to properly distinguish the different stages of the cycle of creation of knowledge and meaning. Usually, a visual and graphical analysis is presented as part of the methodology to deepen the understanding of the different events happening in the one second duration post stimuli window that ideally allows the development of the ability to visually discriminate between CS+ and CS- stimuli, by carefully gazing at the spatio-temporal dynamics in specific movie-frames, where images of, for example, an 8x8 spatial array of ECoG patterns is displayed for different moments of time, for one or more of the following indexes: the pragmatic information index (HRSA, HRAA, HSAA), the original signal amplitude (SA), the analytic amplitude (AA), the analytic phase (AP) and the instantaneous frequency (IF). Sometimes the analytic amplitude signal squared (AA^2) or its logarithm to the base ten (Log10), is also shown as a measure of energy consumption that also illustrates different aspects of the cycle of creation of knowledge and meaning, particularly featuring Null Spikes, signalling important events in brain dynamics.
It is important to note that the pragmatic information indexes HRSA, HRAA and HSAA are derived from another set of indices based on the range of the signal amplitude (RSA), the range of the analytic amplitude (RAA) and the value of the analytic amplitude squared (SAA), also represented as AA^2, where every one of them is divided by a type of Euclidean Distance (ED) between channels, sometimes based on the analytic phase (ED_AP) and others based on the analytic amplitude squared (ED_AA^2).
See also:
Davis, J.J.J. “Pragmatic Information, Intentionality & Consciousness,” Journal of Consciousness Exploration & Research, 9(2), 113-123, 2018. Published in Journal of Consciousness Exploration & Research and Scientific GOD Journal and the preprint is available here: Center for Open Science (OSF)
It is important to note that the pragmatic information indexes HRSA, HRAA and HSAA are derived from another set of indices based on the range of the signal amplitude (RSA), the range of the analytic amplitude (RAA) and the value of the analytic amplitude squared (SAA), also represented as AA^2, where every one of them is divided by a type of Euclidean Distance (ED) between channels, sometimes based on the analytic phase (ED_AP) and others based on the analytic amplitude squared (ED_AA^2).
See also:
Davis, J.J.J. “Pragmatic Information, Intentionality & Consciousness,” Journal of Consciousness Exploration & Research, 9(2), 113-123, 2018. Published in Journal of Consciousness Exploration & Research and Scientific GOD Journal and the preprint is available here: Center for Open Science (OSF)