IEEE Technology and Society Magazine - June 2015 - 34

Components of a BCI
A BCI is a system used to translate electrophysiological
signals, reflecting activity of an individual's central nervous
system, into a user's intended messages that act on the
external world [15]. From an engineering perspective, it
is a communication system, consisting of inputs (user's
neural activity), outputs (external world commands), and
components translating inputs to outputs, known as signal acquisition and signal processing. A high-level block
diagram of a typical BCI is depicted in Fig. 1.
Based on the recording location, BCIs can be divided
into: a) invasive, b) moderately invasive, and c) non-invasive systems. Invasive BCIs involve electrodes or electrode arrays that are directly implanted into the brain
during a surgery. Inside-the-brain implanted BCIs enable
the highest quality measurements of neural activity.
Moderately invasive BCIs, such as electrocorticography (ECoG) are implanted inside the skull, typically on
top of the brain. They provide signals of lower noise and
higher selectivity than non-invasive BCIs, which record
neural signals from the scalp.
Most non-invasive BCIs are based on electroencephalography (EEG). While known to be susceptible to noise and
signal distortion, EEG signals are easily measurable. In addition, EEG-based BCIs have relatively low cost and low risk,
which makes them the most widely used BCI devices [14].
The signal processing component of a BCI typically
consists of two parts: feature extraction and decoding
(translation) algorithms. The feature extraction part
processes recorded neural signals, in order to extract
signal features. These are assumed to reflect specific
aspects of a user's current neural signal. Decoding

algorithms take those abstracted feature vectors and
transform them into application-specific commands.
Depending on the application, many different decoding algorithms are being used in BCIs. As pointed out
in [15], effective decoding algorithms are able to adapt
to: 1) individual user's signal features, 2) spontaneous
variations in recorded signal quality, and 3) adaptive
capacities of the brain (neural plasticity).

Ethical and Legal Considerations
of Neural Engineering

With an increasing number of neural engineering applications, specifically BCIs and neural imaging, researchers
have recognized the need to address emerging ethical
and legal questions [20]-[27]. In 2003, Jonsen introduced
neuroethics as "a discipline that aligns the exploration and
discovery of neurobiological knowledge with human value
system." It was recognized that neuroethics will have to
address questions related to a) incidental findings, b) surrogate and biomarkers of diseases, and c) commercialization of cognitive neuroscience [23].
In 2005, The Committee on Science and Law considered possible legal implications of neural engineering [25].
An emphasis was put on privacy implications of neural
imaging, in particular on the use of neural imaging in
non-medical research. The committee recognized neuromarketing, defined as the field of marketing research that
studies consumers' sensorimotor, cognitive, and affective
response to marketing stimuli [26] and brain fingerprinting, defined as a technique that purports to determine the
truth by detecting information stored in the brain [27], as
emerging non-medical areas using neural imaging data.
The committee observed there are
important similarities between genetBCI System
ic and brain data, in that: 1) "both
genetic and brain data hold out the
Signal Processing
promise of prediction (not only disDigitized
Signal
Feature
Translation
ease, but also behavior)," and 2) "both
Acquisition
Extraction
Algorithm
Signal
types of information expose unique
and personal, and to a large extent,
uncontrollable aspects of a person
that previously were unobservable"
Neuroprosthesis
[25]. Based on these observations, the
Control
committee proposed exploring and
Neuroprosthesis
leveraging for neuroethics those medical, ethical, and legal rules already set
P300 Speller
forth in genetic research.
Control
In [21], Farahany observed that
P300 Speller
modern neuroscience and neural engineering pose an novel set of
Neurogame
Control
legal challenges to the existing U.S.
Neurogame
Self-Incrimination doctrine of the Fifth
Amendment, which states that "no
Figure 1. High-level block diagram of a typical brain-computer interface.
person shall be compelled to prove a

IEEE Technology and Society Magazine

∕

june 2015

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - June 2015