Intention concepts and brain-machine interfacing

Abstract

Intentions, including their temporal properties and semantic content, are receiving increased attention, and neuroscientific studies in humans vary with respect to the topography of intention-related neural responses. This may reflect the fact that the kind of intentions investigated in one study may not be exactly the same kind investigated in the other. Fine-grained intention taxonomies developed in the philosophy of mind may be useful to identify the neural correlates of well-defined types of intentions, as well as to disentangle them from other related mental states, such as mere urges to perform an action. Intention-related neural signals may be exploited by brain-machine interfaces (BMIs) that are currently being developed to restore speech and motor control in paralyzed patients. Such BMI devices record the brain activity of the agent, interpret ("decode") the agent's intended action, and send the corresponding execution command to an artificial effector system, e.g., a computer cursor or a robotic arm. In the present paper, we evaluate the potential of intention concepts from philosophy of mind to improve the performance and safety of BMIs based on higher-order, intention-related control signals. To this end, we address the distinction between future-, present-directed, and motor intentions, as well as the organization of intentions in time, specifically to what extent it is sequential or hierarchical. This has consequences as to whether these different types of intentions can be expected to occur simultaneously or not. We further illustrate how it may be useful or even necessary to distinguish types of intentions exposited in philosophy, including yes- vs. no-intentions and oblique vs. direct intentions, to accurately decode the agent's intentions from neural signals in practical BMI applications.

Keywords: BCI; BMI; action intention; intentional; philosophy of mind.

Figure 1

An overview of cortical responses reported in recent functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) studies that explicitly aimed at investigating intentions in healthy subjects. Peaks are plotted on a standard brain from SPM8 on (A) the left and (B) right hemisphere. The approximate locations of the prefrontal and the inferior parietal cortex are indicated in yellow and blue, respectively. Reported intention-related peaks in both hemispheres exhibit a widespread spatial distribution across the frontal, parietal, occipital, and temporal lobes.

Figure 2

The working principle and current approaches of BMI. In (A), BMI-based control of a prosthetic hand by using recordings of brain activity is depicted. The projections from the motor cortex (1) constituting the cortico-spinal pathways (2) may be disrupted, e.g., by spinal-cord injury (3). Then, suitable electrodes (4) can be used to record persistent motor-cortical activity, which is transmitted by a technical connection (5; either wire-based or wireless) to a decoder (6) extracting control signals for an external actuator (7). If the primary motor cortex is destroyed, such as due to stroke, cognitive control signals may still be recorded from alternative areas such as the prefrontal cortex (8). As summarized in (B), neural control signals may thus range from low-level motor signals, such as related to movement direction or velocity, to more high-level cognitive signals related to abstract action goals, subjective preferences, and intentions. Output signals may be used to restore movement (e.g., of an external actuator) or communication. The input-output mapping can be realized in a direct way, e.g., if right- and leftward movement-related neural activity controls the respective right- and leftward movements of an effector, the intention to grasp a cup is directly translated into the corresponding grasping action, or the intention to say the word “hello” is directly translated into speech. Indirect strategies would be, for example: using imagined leg vs. tongue movements to control right- vs. leftward movements of a robotic arm.

Figure 3

(A) Schematic representation of sequential (A) vs. hierarchical (B) vs. (C) mixed models of intention organization in time. Three different scenarios are depicted for the case of a threefold intention concept, roughly corresponding to the F-, P-, and M- intentions proposed by Pacherie (2006). The durations of F-, P-, and M-intentions are depicted by colored bars. A mixture of hierarchical and sequential relations is shown in (C). Which of these different scenarios is true in a given situation would have important consequences for attempts to decode intentions; for example, the detection of an F-intention would rule out the simultaneous presence of the corresponding P- and M-intentions in the purely sequential (A) but not in the hierarchical (B) model. Note that Pacherie’s concept favors (C), particularly in her recent work (Pacherie, 2008).