Featured Posts
Recent Posts
Search By Tags
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square

Methodology of Consciousness Research (Consciousness Series: Part 3)

Sometimes in research, a good way to study how a system normally functions is to study models in which something is wrong with the system. This is why in order to study consciousness, patients with disorders of consciousness are often studied in comparison to healthy volunteers using neuroimaging. Studying people with disorders of consciousness is especially important, as we will discuss in the next blog post, because of the monumental ways in which BCI can transform healthcare and overall lifestyle for people experiencing disorders of consciousness. Many people with disorders of consciousness, especially those in a vegetative state, were previously considered to be wholly unconscious. However, thanks to cutting-edge research, we have begun to realize that some of these patients, depending on the level of brain trauma, may be completely mentally conscious but simply unable to physically respond. Last blog post, we covered some of the basic research methods of studying consciousness, but today, we will be diving more deeply into examples of research studies in consciousness, particularly with patients with disorders of consciousness. Without further ado, let's get right into it.

Early Studies of Disorders of Consciousness: Problems Assessing Awareness:

Earlier designs of studies of behaviorally unresponsive patients oftentimes used passive stimuli to study patterns of response in the brain via PET scan, fMRI, and EEG. For example, one of the earliest studies (de Jong, 1997) studied regional cerebral blood flow in a post-traumatic vegetative patient. The patient's mother reading a story to them served as the passive stimuli, and the corresponding response in the brain was astonishing. For a patient who was thought to be completely unresponsive and unreceptive, they showed brain activation in the anterior cingulate and temporal cortices, which are areas involved in emotional processing. This result supposedly showed that the patient was processing the tone and contents of his mother’s auditory stimulus. Another early study in 1998 conducted PET scans on vegetative patients using a visual stimulus of faces that were familiar to each individual patient, which resulted in activity in the right fusiform gyrus, a brain region linked with recognizing human faces. Roughly a decade later, another study (Di, 2007) used event-related fMRI to study auditory responses to the patients’ names being said by a person familiar to them. The study sample included 7 vegetative and 4 minimally conscious patients. Results showed that all minimally conscious patients as well as two vegetative patients had higher-order associative temporal lobe activation, which indicates primary auditory perception. Interestingly, the two vegetative patients who showed these results were also the only two to improve over time to minimally conscious states. However, one limitation of this study is that there was no healthy, control group as comparison. Therefore, the response could have simply been a low-level, automatic brain response to general auditory or emotional stimuli rather than a specific brain response to the patient’s name, which would be more indicative of consciousness. We will see later that this is one specific area in consciousness research design that has been significantly improved in modern paradigms assessing consciousness. One early study that avoided this issue (Staffen, 2006) instead used two strings of speech, one containing the patient’s name at one point and one containing a random name, iterated by unfamiliar voices to vegetative patients. The purpose of this design was to more specifically link responses to patient name. The event-related fMRI showed differential cortical processing in a part of the medial prefrontal cortex, which is important in social cognition and other higher-order memory and decision making functions; this response similarly occurred in 3 healthy volunteers. However, one important point the authors made is that reacting to your name is one of the most basic and automatic forms of language processing, so it is not necessarily linked with higher levels of comprehension / consciousness. One of the largest of these earlier studies using passive stimuli (Coleman, 2009) studied 41 patients with disorders of consciousness using fMRI and language paradigms that

increased in complexity. Roughly 50% of the patients showed fairly normal or normal responses in the temporal lobe to lower complexity auditory tasks, such as auditory or speech contract and sound perception. Four patients, however, also showed normal fMRI activity for high-level speech comprehension, which shows these patients still have linguistic processing even through they are behaviorally diagnosed as vegetative or minimally conscious. This idea will be further dissected for its context and importance in the next blog post. While PET scans and fMRI are popular neuroimaging tools for nonresponsive patients, another important tool is EEG. EEG is sometimes even preferred in studying disorders of consciousness since it is cost-effective and mobile versus fMRI. For example, EEG can be used during the acute period of brain-injured patients and can help make some broad generalizations about any sort of brain death and/or cortical damage. However, it is difficult to assess awareness using EEG due to the difficulty of finding a resting state signature. A resting state signature means the baseline EEG activity and frequencies exhibited by a patient with a disorder of conscious absent of any stimuli. Studying healthy volunteers has shown that fast activity that is low voltage is a typical EEG signature of wakefulness, while slow activity that is high voltage is a typical EEG signature of NREM sleep. Despite these discoveries though, it is still difficult to assess resting state signatures in patients with disorders of consciousness due to their often abnormal raw EEG patterns.

As seen in the last paragraph, it was difficult in these earlier studies to construct paradigms that avoid eliciting automatic responses and thus definitively require conscious attention and processing. The main question in consciousness research design is whether or not what is being measured is an unconscious reflex or a conscious, thought-out response. For example, let’s see another early anesthetic study that struggled with this issue. Similar to patients with disorders of consciousness, people under anesthesia have become a popular model for studying consciousness using neuroimaging due to the ability to actively manipulate consciousness. The anesthetic study (Davis, 2007) used fMRI to observe nonsedated, lightly sedated, and deeply sedated healthy volunteers, exposing them to the same speech stimuli as a control factor. The speech stimuli were sentences with either no ambiguous words, signal-correlated noise, or sentences with ambiguous words. In response to the first two stimuli, healthy volunteers showed a brain response in their temporal lobes at all levels of sedation, which is thus not a reliable biomarker for awareness since it occurred in the same normal way at all three sedation levels. However, the last stimulus showed light and deep sedation patients to have no frontal lobe and posterior temporal lobe activation, which are neural correlates of semantic processing, similar to the nonsedated group. In other words, at all levels of sedation, none of the participants showed semantic processing which makes sense since this stimuli included sentences with ambiguous words. While this is an exciting conclusion, you cannot always draw this conclusion; this is because it is synonymous with saying “well, this neural activity is often correlated with this state of being; therefore, everytime we see that neural activity then that person is in that correlated state of being." That is like saying, for example, everytime we see amygdala activation the person is in a state of fear, which we know to not be true.

Recent Studies of Disorders of Consciousness: Moving the Field Forward by Decoding Neuroimaging Signatures:

Now that we have discussed earlier studies in which findings are not as reliable due to the possibility of unconscious, neural reflexes at play, we will move on to discuss more recent studies focusing on brain activity directly related to conscious choices and responses, which are indicative of neural correlates of consciousness. For example, one study (Haynes 2007) asked a sample of healthy volunteers to mentally decide between and perform one of two tasks (either adding or subtracting two numbers). The experimenter would then observe the brain patterns and subsequently use this fMRI signature to properly decode 80% of trials to confirm which task the participant chose based solely on observing their brain activity. This demonstrates that thoughts have specific brain activation pathways and patterns that can be detectable and thus usable and predictable. Similarly, another study (Gallivan 2011) used the same principle as Haynes but to decode the movement a person was about to perform before they even performed it. You can probably guess that decoding such neuroimagery signatures are important to BCI, which we will discuss later.

Obviously, these study designs are different from the ones described earlier in this section since these are active thought processes and decisions that participants are making, as opposed to passive reception. This type of research design has become the norm for consciousness research because it allows researchers to then determine awareness even in the absence of behavioral responses, such as in disorders of consciousness. This stronger research design is based on the foundation of studying healthy participants who imagine performing a particular task so that reliable brain patterns and fMRI / EEG signatures can be developed for use in such studies of patients with disorders of consciousness as a comparison. Concrete examples of neuroimagery signatures, or neural correlates of consciousness, are motor and premotor cortices activity while a participant imagines squeezing or moving their hands (Jeannerod & Frak 1999), or parahippocampal gyrus and posterior parietal cortex activity (associated with spatial navigation) while a participant imagines moving between different locations (Aguirre et al. 1996). Another clear example of this is in the Boly 2007 study, in which experimenters asked healthy participants to imagine hitting a tennis ball with a coach in response to the word “tennis” and to imagine walking between rooms in their own homes in response to the word “house.” In response, participants exhibited supplementary motor area (rigorous motor) activity for the prior task and parahippocampal cortices, posterior parietal lobe, and lateral premotor cortices (spatial navigation) activity for the latter task. As stated earlier, this information can then be used in BCI to predict intentions and behaviors.

Recent Studies of Disorders of Consciousness: Assessing Awareness Based on Neuroimagery Signatures:

The previous paragraph describes the first part of consciousness research design: using healthy volunteers to determine and decode neural correlates of consciousness and neuroimagery signatures. Now let’s talk about the next part: using those signatures to determine awareness in behaviorally unresponsive patients with disorders of consciousness. For example, one study (Owen 2006, 2007) showed that a clinically diagnosed vegetative state patient, who had been unresponsive for 6 months following a car crash and prior to the study, exhibited the same fMRI brain response as those healthy participants in the Boly 2007 study (remember: "tennis" vs. "house" task). This proved that they were conscious and aware of their surroundings, even though they could not speak or move. Another study (Monty 2010) replicated this same response from the Boly 2007 study in 4 of 23 vegetative state patients using fMRI (17%). This has led to some controversy since the diagnosis from vegetative to minimally conscious medically requires the raising of a hand or finger in response to verbal commands, but if someone shows neural intentions of doing so in response to verbal command, should their diagnosis change in the same manner? The key is replication; skeptics say brain imaging is less reliable than motor response, but coincidences can occur in both cases. Hence, replication is extremely important; this is why replicated neuroimaging signatures of consciousness in response to specific tasks are becoming more robust and reliable in consciousness research. Other skeptics have proposed that the responses to words, such as those in Boly and Monti’s studies may be an example of an “implicit preconscious neural response;” however, to argue against this criticism is the fact that all of the participants had hemodynamic responses that were sustained for at least 30 seconds until the experimenter indicated to switch tasks, in which case the neural activity subsequently changed. Plus, the responses were in pre-studied regions known to respond to these imaginary tasks, not in regions involved with simple automatic word processing.

Other paradigms used to assess awareness, besides using single command tasks such as imagining playing tennis, are those that require participants to adapt different mindsets depending on the task. For example, in one study (Monti 2009), healthy participants were asked to listen to a string of neutral words and count how many times a certain word was repeated, which activated the frontoparietal network, a brain region known to be associated with target detection and working memory. After testing healthy participants to decode the neuroimagery signature for this type of task, they then tested a brain-injured patient and were able to elicit the same brain pattern both for a sustained amount of time throughout the task and depending on the task condition and what adaptive mindset was required (ie - what word was asked to keep count of). Monti then used the same patient to perform another method that year in which he presented pictures of faces and houses and asked the participant to switch focus between the two superimposed images. In healthy participants, this shift from face to house continuously resulted in a time-locked alteration in neural activity between the fusiform gyrus to the parahippocampal gyrus. This continuous alteration in fMRI activity creates a distinct pattern, and since the stimulus is constant throughout the pattern, it can be attributed to conscious, aware, and willful decisions to switch focus rather than automatic responses that are unreliable due to the unpredictability of the external stimulus.

Similar to fMRI, EEG is also a common tool in consciousness research for developing neuroimaging signatures in order to assess awareness in patients with disorders of consciousness. For example, a 2008 study by Schnakers used a similar task to the one Monti used in the last paragraph in which minimally conscious patients listened to a string of verbal stimuli and had to count the number of times their name was mentioned. Visible and reliably large P3 components were seen in the EEG, which is a characteristic EEG signature of target detection, proving that these patients could follow commands and were therefore aware and conscious. Other earlier studies (Cincotti 2003, Wolpaw 1991) also set a precedent by discovering that motor imagery shows an EEG signature of sensorimotor activity, which is similar to the signature seen during actual execution of movements. This therefore became a common test for awareness and consciousness in patients with severe brain injuries. Interestingly, one can also use EEG signatures to determine the type of motor imagery someone

is mentally envisioning. One of the first studies to do this (Kotchoubey 2003) used a locked-in patient who had a resting state of slow EEG activity. However, when asked to raise the left hand or right hand, there was a distinct shift in EEG activity, with the motor imagery (ie - imagined movements) showing up as reductions in EEG power (also known as event-related desynchronizations or ERDs) in the mu and / or beta bands over the correlating neural region. In simpler terms, EEG power is the overall voltage from your brain’s electrical signal, which stems from your neuron’s firing. Thus, a reduction in EEG power is a reduction in voltage, which during this task would occur in specific frequency ranges of brain waves (in this case the mu and/or beta bands, associated with voluntary movements and consciousness/alertness respectively) over the lateral premotor cortex, which is associated with hand movements (Pfurtscheller & Neuper (1997). Similar to ERDs, there are also ERS’s (event-related synchronization), which include increases in EEG power; ERS’s often occur around the motor areas exhibiting ERDs during motor imagery tasks. Another example of predicting types of motor imagery occurred in the Guger 2003 study. During this study, experimenters were able to, with 60-91% accuracy, decode which of two tasks (squeezing right hand or squeezing toes) that healthy individuals were mentally performing. Thus, patients who are behaviorally unresponsive but show these same EEG patterns during such a task are showing complex cognitive functions, including sustained attention, language comprehension, response selection, and working memory, that are attributed to normal consciousness and awareness. Although some skeptics claim that these neural reactions may be automatic in response to the words “right hand” or “toes,” when patients were asked to not follow the commands / generate motor imagery, they had no reaction to the commands, demonstrating that they were able to comprehend and follow tasks. Again, the key to meeting such skepticism in consciousness research is consistency and replication of results. However, not all disorders of consciousness can be researched and regarded in the same way. One great example of this is the Cruse 2012 study. In this study, non-traumatically injured patients in minimally conscious states had positive EEG responses matching their behavioral responses to the same tasks, proving that neuroimagery can be used in the same way behavioral responses are used in order to assess awareness. Likewise, when they ran the same test on traumatically injured patients, who were unable to be behaviorally responsive, within the study, only 75% showed the same positive EEG responses (75%), which is indicative of consciousness and awareness. This goes to show that it is important to take into consideration the neuropathological and behavioral differences in patients suffering from traumatic versus non-traumatic brain injury. Plus, besides behavioral differences, neuroimaging results between the two groups can be vastly different for other more internal, neurological reasons. For example, traumatic brain injury often causes diffuse axonal injury, which affects both hemispheres and many other structures of the brain and may show up in neuroimaging.

Applying the Same Principles to Anesthetic Studies:

Besides using fMRI and EEG to study disorders of consciousness, it can also be used in anesthetic studies. The Adapa 2011 study used fMRI and the “imagine playing tennis” task, which is commonly used on minimally conscious and vegetative patients, on healthy volunteers in a nonsedated, lightly sedated, and deeply sedated state. Only the deeply sedated volunteers had no responses at all to the task, proving they were unconscious. Once each group was awake again and fully recovered from the propofol, fMRI activity was once again normally observed in the premotor cortex in response to the task, showing that it is a reliable test for consciousness and uncosnciousness alike (ie - presence or absence of brain activity through neuroimaging).

Connections to BCI:

Now that we have extensively discussed examples of consciousness research design and dove deeper into methodology, you may be thinking how BCI can tie into all of this and provide a concrete application for all this research. BCI, also known as “thought-translation devices,” allow patients with disorders of consciousness to communicate with the outside world based on their brain activity. This brain activity has been deciphered, as you can see, from a plethora of consciousness research. Both fMRI and EEG alike can be used in such BCI. For example, the Sorger 2009 study had participants follow a specific procedure based on the one of the four available answer choices they choose, which then allowed a machine to decode their answer choices with a 94.9% accuracy through analysis of the blood-oxygen-level-dependent responses. Similarly, another study (Owen / Coleman, 2008) developed a task by which fMRI and known patterns were used to distinguish "yes" from "no" responses to simple questions posed by experimenters. In their case, imagining playing tennis was associated with a "yes" response, and imagining walking around your own house was associated with a "no" response. This developed paradigm was later used by Monti in 2010 with 16 healthy volunteers; as a result, Monti was able to successfully decode "yes" from "no" responses to three questions per participant with 100% (!!!!!!) accuracy due to the changes in activity in the supplementary motor area for "yes" responses or parahippocampal place area for "no" responses. Monti then used this same technique with a traumatic brain injury patient, who was diagnosed as vegetative for 5 years but was still able to produce such fMRI "yes" or "no" responses to specific questions about where they were 5 years ago before the accident, their father’s name, etc. This led to their diagnosis being changed to minimally conscious. Clearly, such fMRI BCI can use neuroimaging signatures to allow patients to simply and reliably respond to questions posed by people in their environment. These BCI are also adaptable to the person; Weiskopf (2004) taught four participants how to self-regulate their own fMRI signal by differing their visual, mental imagery or by performing other regulatory strategies such as dancing, clenching, spatial navigation, etc.

The use of EEG in the development of BCI has paved paths for some more complex communication skills for patients with disorders of consciousness. Specifically, the Farwell & Donchin study in 1988 produced the P300 modulation paradigm for such advanced communication. A matrix of the alphabet with a flash across the letters was presented to severely paralyzed or locked-in patients in pseudorandomized order. The letter that flashes right before an evoked P300 component (associated with decision-making) is the letter that the patient is attempting to communicate. However, this modulation would not work well for vegetative or minimally conscious patients who cannot perform the visual fixation required for that BCI. So, Sellers & Donchin modified the task to work with 4 visual or auditory stimuli associated with either a "yes", "no", "pass", or "end" response. This allowed such patients to effectively communicate, although the auditory stimuli were less accurate and reliable responses. Such research importantly leads to the development of reliable BCI devices that allow patients with specific mental state classifications to produce external responses and hone in on their communication capabilities.

Note from the Blogger:

Thanks for joining me as we essentially went through a short lit review of past consciousness research studies. Join me next week, when we will step back and look more at the broader clinical basis of disorders of consciousness, which are a popular area of research within the greater field of consciousness research.


  • https://courses.lumenlearning.com/boundless-psychology/chapter/introduction-to-consciousness/

  • https://webcache.googleusercontent.com/search?q=cache:DRbR4Pat9ggJ:https://www.academia.edu/1967589/Cognition_in_the_Vegetative_State+&cd=2&hl=en&ct=clnk&gl=us

  • https://en.wikipedia.org/wiki/Functional_magnetic_resonance_imaging

  • https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5641373/

  • https://www.psychologytoday.com/us/blog/consciousness-self-organization-and-neuroscience/201702/what-the-heck-is-claustrum

  • https://www.uwo.ca/bmi/owenlab/pdf/2013-Owen,%20Annual%20Review%20-%20Detecting%20consciousness%20A%20role%20for%20neuroimaging.pdf

  • https://plato.stanford.edu/entries/consciousness-neuroscience/

  • https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4423300/

  • file:///home/chronos/u-cc5ef590d908de473ab97803e84584fdccb5a36f/MyFiles/Downloads/Frohlich%20et%20al.%202019%20(2).pdf

  • Black Instagram Icon