fMRI Neurofeedback | 2021
Implicit decoded neurofeedback training as a clinical tool
Abstract
This chapter will concentrate on the implicit nature of Decoded Neurofeedback (DecNef), which can serve as a potential alternative to exposure therapy or counter conditioning for alleviating fear and anxiety. DecNef has been shown to be successful in reducing fear responses to specific stimuli by inducing neural representations of feared stimuli without inducing conscious awareness of the stimuli. Such implicit DecNef may be suitable for the treatment of PTSD, phobia and other anxiety disorders, especially when patients are intolerant to aversive conscious exposure to their feared stimuli during conventional therapies. Our hope is not to replace the existing clinical practice with DecNef training, but rather to make some additional treatment options available to patients so that they have more flexibility in choosing how they would like to overcome their issues with fear and anxiety. Implicit nature of Decoded neurofeedback (DecNef) Decoded Neurofeedback (DecNef) is a relatively novel type of neurofeedback which allows real-time neurofeedback based on spatial activation patterns in targeted brain areas (Watanabe et al., 2018). Combined with fMRI multivoxel decoding techniques to dissociate different neural representations (e.g.,(Yamashita et al., 2008)), DecNef provides feedback regarding the likelihood that a certain representation is activated in a given brain area. For example, it provides feedback to a participant based on whether his or her visual cortex is likely to be representing a red rather than a green target stimulus (Amano et al., 2016, Koizumi et al., 2016) or whether their prefrontal areas are likely to be representing a high rather than a low state of perceptual confidence (Cortese et al., 2017). One of the features which makes DecNef distinguishable from many other neurofeedback techniques is its implicitness (Watanabe et al., 2018). That is, participants remain unaware of what neural representation is being induced throughout DecNef training sessions (Figure 1). They are generally instructed to somehow manipulate their brain activation in order to receive better feedback (e.g., a circle size on a screen) through trialand-error. Yet, they are never provided with any explicit instruction e.g., to imagine a red stimulus. Despite the fact that participants generally learn to successfully induce the targeted brain activation patterns, post-training questionnaires and/or forced-choice questions have consistently revealed that participants do remain unaware of the identity of targeted neural activation patterns (Amano et al., 2016, Cortese et al., 2017, Koizumi et al., 2016, Shibata et al., 2011, Taschereau-Dumouchel et al., 2018)((Shibata et al., 2019) for review). The fact that participants remain unaware of the induced neural representations may sound surprising to some. However, it is in line with a large body of literature on consciousness: Neural representations in confined brain areas need to be connected with wider parts of the brain to reach conscious awareness (Brown et al., 2019), although the exact mechanisms involved in consciousness remain a matter of active debate. Previous studies with DecNef have shown that the neural representation induced in a targeted brain area is highly localized (Shibata et al., 2011). For example, even when an intended stimulus representation of a right tilted grating is successfully induced in the early visual cortex targeted during DecNef training, activation patterns in other brain areas often fail to corepresent the right tilted grating (Shibata et al., 2011). Because the induced representation is generally confined to the targeted brain area, this representation may remain outside of conscious awareness. Figure 1. Schematics of the implicit DecNef training to help overcome fear and anxiety. DecNef may benefit the treatment of fear-related disorders with its implicit nature Its implicit nature makes DecNef a unique tool to potentially treat anxiety disorders such as post-traumatic stress disorder and phobias. One of the common treatments for anxiety disorders is exposure therapy. During exposure therapy, a patient often experiences strong aversive emotions because he or she needs to observe or imagine the scenes and objects related to their disorders, e.g., an image of a car that reminds them of a traumatic car accident. Exposure therapy is thought to rely on a fear extinction process through which a participant eventually learns that the feared objects are no longer associated with pain or trauma. Despite its effectiveness, the distress of therapy leads to a non-negligible rate of dropouts, with an estimated range from 0 to as high as 70 % (Loerinc et al., 2015, Zayfert et al., 2005). DecNef, with its implicitness, may help overcome such issue of distress and benefit the treatment of anxiety disorders. Specifically, instead of explicitly viewing or imagining a feared object, a patient could unknowingly induce a neural representation of the feared object through DecNef training. Repeated implicit reactivation of the neural representation of a feared object, when paired with reward as described below, may eventually alleviate fear towards the object in a manner similar to explicit exposure therapy. Here we describe a few studies that have directly examined this possibility. Progressive development of DecNef as a clinical tool The first study examined the effectiveness of DecNef for reducing fear-like responses among healthy participants (Figure 2. (Koizumi et al., 2016)). In this study, participants initially underwent a fear conditioning session to obtain fear-like response to two colored stimuli, red and green gratings, both of which were paired with uncomfortable electric shocks. Participants then went through a three-day DecNef session in which they were provided with rewarding feedback when their early visual cortices (V1/V2) were more likely to be representing either the red or green stimulus. Which color served as their target stimulus was counterbalanced across participants and was fixed throughout the session. Better induction of the activation patterns representing the target stimulus was followed by a larger disc presented on the monitor, whose size was proportional to the actual monetary reward provided to the participants at the end of each day. After the three-day DecNef session, participants underwent a test session in which they were explicitly presented with the fear conditioned red and green stimuli. In the test session, participants showed reduced skin conductance responses and amygdala activation to the stimulus that served as target and had its neural representation paired with reward during the DecNef training, compared to the control stimulus whose neural representation was not paired with reward during DecNef training. Importantly, when participants were asked to guess which color, red or green, served as their target neural representation after completing all the sessions, they could only guess the target color at chance level. Moreover, fear-like responses to the neurally induced activation patterns of feared objects during the DecNef session were significantly smaller than the responses observed when the stimuli were explicitly presented. These results suggest that DecNef is capable of reducing fear-like responses to feared objects without explicit and aversive exposure. Figure 2. Schematics of the procedure in Koizumi et al. (2016) (Koizumi et al., 2016). In this study, healthy participants went through a fear conditioning session to form two fear memories to red and green colored gratings. They then underwent DecNef training (with grey gratings) to reduce one of these fear memories. In this example, the red memory is targeted during training, and the fear response to the red stimulus is reduced relative to the control (green) stimulus. One limitation of the study by Koizumi et al. (Koizumi et al., 2016) was that, prior to the threat-conditioning session, the red and green stimuli were explicitly presented to the participants to enable the multivoxel pattern decoding of their neural representations. This explicit decoding session was not unpleasant to the participants as it was prior to the formation of aversive memory for the decoded representations. However, if the procedure were to be applied to the actual treatment of anxiety disorders, aversive exposure to the feared objects for the sake of decoding would undermine the key advantage of this approach: that is, the ability to reduce fear associations via the implicit and non-aversive nature of DecNef training. To make the entire procedure implicit, it is necessary to also render the decoding process implicit. To achieve this, Taschereau-Dumouchel et al. (Taschereau-Dumouchel et al., 2018) has elaborated the aforementioned DecNef training procedure by integrating another novel technique, hyperalignment (Guntupalli et al., 2016, Haxby et al., 2011). With hyperalignment, this study aimed to alleviate fear-like responses to feared animals among sub-clinically phobic participants. Briefly, hyperalignment enabled the construction of a decoder for the representation of a feared animal in each participant’s brain that was based on, or “borrowed from”, the brain activation patterns of surrogate participants. The first stage of the study involved learning how to align the neural activation patterns of surrogate participants to the activation patterns of a participant with a subclinical level of phobia. This first stage was achieved using activation patterns obtained while the surrogates as well as phobic participants explicitly observed various images of animals that did not include the feared animals. For example, if the phobic participant is fearful of both snakes and spiders, the first stage used the activation patterns in ventral temporal cortex for various animals such as butterflies, dolphins, etc., but did not involve those for snakes or spiders. Data from this stage allowed co