Optical Communication among Oscillatory Reactions and Photo-Excitable Systems: UV and Visible Radiation Can Synchronize Artificial Neuron Models

Article abstract of DOI:10.1002/anie.201702289

Neuromorphic engineering promises to have a revolutionary impact in our societies. A strategy to develop artificial neurons (ANs) is to use oscillatory and excitable chemical systems. Herein, we use UV and visible radiation as both excitatory and inhibitory signals for the communication among oscillatory reactions, such as the Belousov–Zhabotinsky and the chemiluminescent Orban transformations, and photo-excitable photochromic and fluorescent species. We present the experimental results and the simulations regarding pairs of ANs communicating by either one or two optical signals, and triads of ANs arranged in both feed-forward and recurrent networks. We find that the ANs, powered chemically and/or by the energy of electromagnetic radiation, can give rise to the emergent properties of in-phase, out-of-phase, anti-phase synchronizations and phase-locking, dynamically mimicking the communication among real neurons.

Full text of DOI:10.1002/anie.201702289

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201702289
German Edition: DOI: 10.1002/ange.201702289
Nonequilibrium Processes
Optical Communication among Oscillatory Reactions and Photo-
Excitable Systems: UV and Visible Radiation Can Synchronize
Artificial Neuron Models
Pier Luigi Gentili,* Maria Sole Giubila, Raimondo Germani, Aldo Romani, Andrea Nicoziani,
Anna Spalletti, and B. Mark Heron
Abstract: Neuromorphic engineering promises to have a rev-
olutionary impact in our societies. A strategy to develop
artificial neurons (ANs) is to use oscillatory and excitable
chemical systems. Herein, we use UV and visible radiation as
both excitatory and inhibitory signals for the communication
among oscillatory reactions, such as the Belousov–Zhabotin-
sky and the chemiluminescent Orban transformations, and
photo-excitable photochromic and fluorescent species. We
present the experimental results and the simulations regarding
pairs of ANs communicating by either one or two optical
signals, and triads of ANs arranged in both feed-forward and
recurrent networks. We find that the ANs, powered chemically
and/or by the energy of electromagnetic radiation, can give rise
to the emergent properties of in-phase, out-of-phase, anti-phase
synchronizations and phase-locking, dynamically mimicking
the communication among real neurons.
puters can be implemented by using either conventional
[
5]
passive and active circuit elements, or two-terminal devices
with a non-volatile adjustable internal state such as mem-
[
6]
ristive components, or oscillatory and excitable chemical
[
7]
[8]
reactions. In this work, we use oscillatory reactions (Osc)
and excitable photochromic or luminescent species (Exc) as
artificial neuron (AN) models and we study their communi-
cation through UV/visible radiation.
Neurons are nonlinear dynamic systems that, usually,
[
9]
synchronize when they communicate.
Therefore, to
approach the computational capabilities of the human brain,
it is useful to implement and study the synchronization of
nonlinear chemical systems. Nonlinear chemical systems have
[
10]
been typically coupled through continuous mass exchange,
[11]
electrochemical linkage,
or mechanical/light-induced
[
12]
pulsed release of chemicals. Herein, we propose the use
of UV/visible radiation as excitatory and inhibitory optical
signals that allow us to establish in-phase, anti-phase, or out-
of-phase synchronizations and phase-locking. The results of
experiments and simulations are presented, regarding the
optical communication between two ANs and either one or
two optical signals (S), and between three ANs organized in
either feed-forward (FFN) or recurrent (RN) networks.
The reactions and compounds used in this research are
listed in Scheme 1.
T
he human brain and intelligence are drawing attention
because their operation is inspirational in various disciplines
[1]
that deal with complexity. Chemists, in collaboration with
colleagues of other disciplines, are contributing to the
[2,3]
development of neuromorphic engineering. Neuromorphic
engineering implements surrogates of neurons through non-
[
4]
biological systems either for neuro-prosthesis or to devise
brain-like computing machines. Brain-like computing
machines will exhibit the peculiar performances of human
intelligence, such as learning, recognizing variable patterns
and computing with words as some programs have com-
menced to do. However, it is expected that brain-like
computers will have the advantage of requiring much less
power and occupying much less space than our best electronic
supercomputers. Surrogates of neurons for brain-like com-
The communication architectures (a, b, g), which have
been devised, are depicted in Scheme 2. In a, the transmitter
and the receiver of the communication are in the same cuvette
and in the same phase, possibly with one component chemi-
cally protected by micelles. In b, the transmitter and the
receiver are in the same cuvette but in two immiscible phases
(water–ionic liquid). In g, the transmitter(s) and receiver(s)
are in two different cuvettes. The networks have been
attained by hybridizing or upgrading the a, b, and g
architectures.
[*] Prof. P. L. Gentili, M. S. Giubila, Prof. R. Germani, Prof. A. Romani,
A. Nicoziani, Prof. A. Spalletti
Three examples of communication established between
two ANs by one optical signal are shown in Figure 1. In all the
situations, the transmitter is in an oscillatory regime simulat-
ing a pacemaker neuron. The transmitter sends an excita-
tory optical signal (indicated by the arrow) to the receiver.
When the receiver is photo-excitable, it synchronizes with the
transmitter either in phase or out of phase, depending on its
relaxation time (t). If t is short, the synchronization is in
phase (Df = 0), whereas if t is long it is out of phase (Df > 0).
For example, in the left panels of Figure 1, the results of the
Department of Chemistry, Biology and Biotechnology
University of Perugia
Via Elce di sotto 8, 06123 Perugia (Italy)
E-mail: pierluigi.gentili@unipg.it
[
9]
Prof. B. M. Heron
Department of Chemical Sciences, School of Applied Science,
University of Huddersfield, Queensgate
Huddersfield, HD1 3DH (UK)
Supporting information, including descriptions of materials, facili-
ties, models for the simulations, and further experimental data, and
the ORCID identification number(s) for the author(s) of this article
can be found under:
communication between BZ(Ce) and the fluorophore F are
1
https://doi.org/10.1002/anie.201702289.
depicted. The transmitter and receiver are in the same
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
Scheme 2. Three communication architectures: 1-cuvette–1-phase (a),
-cuvette–2-phases (b), and 2-cuvettes (g).
1
communication is demonstrated by the synchronized periodic
F green fluorescence (Figure 1a,c). The synchronization is
perfectly in phase because the relaxation time of the emission
from F lasts only a few nanoseconds. The inset of Figure 1c
shows that F fluorescence oscillates, but decreases too, due to
an incomplete oxidative protection carried out by micellar
aggregates (see also the Supporting Information). The same
kind of master-and-slave relationship can be achieved when
the receiver is a photo-excitable T-type photochromic species
like P (see the middle panel in Figure 1). However, the
synchronization is out of phase because the P relaxation time
Scheme 1. Oscillatory and photochromic reactions and fluorophore F.
cuvette, but the F molecules are protected within SDS
micelles from the harsh chemical environment of the BZ
+
4
+3
reaction. The recurrent transformation of Ce to Ce and
+
3
+4
Ce to Ce determines significant transmittance changes in
the UV (Figure 1b). A continuous UV radiation going
through BZ(Ce) (the master) is transformed in a periodic
signal that is absorbed by F (the slave). The success of the
is long, being tens of seconds. In Figure 1, BZ(Ce) transmits
2
Figure 1. Schemes and outputs of three 2ANs-1S communications. Left: BZ(Ce) !F; middle: BZ(Ce) !P; right: BZ(Ce) !BZ(Ru).
1
2
1
2
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
a periodic UV signal to P (in either architecture b or g). The P
tion. When the signals are unidirectional we have FFNs. Two
color band centered at 463 nm oscillates (Figure 1d) in
synchronization, with a phase lag of circa 508 (Figure 1e), in
agreement with the simulations (Figure 1 f and the Support-
ing Information). When the receiver is in an oscillatory
regime, as in the case shown on the right panel of Figure 1,
which involves the photo-sensitive BZ reaction with [Ru-
examples of FFNs are shown in Figure 3. FFN is a hybrid of b
1
and g architectures. In fact, OscBZ(Ce) transmits an excita-
tory periodic UV signal to Exc(P), dissolved in the IL and
placed in the same cuvette (b). P synchronizes almost in phase
with BZ(Ce) (Figure 3a). Then, P transmits a periodic blue
2
light that excites F contained in another cuvette (g). The
green light emitted by F is perfectly in phase with the blue
2
+
(
bpy)3] as the catalyst [BZ(Ru)] (Figure 1g), the success of
the communication is demonstrated by the achievement of
a phase-locking condition (that is, in one oscillation of the
sender, an integral number of receiver oscillations occur).
intensity transmitted by P (Figure 3b). Therefore, FFN is
1
analogous to the officer–non-commissioned officer–soldier
relationship because the message sent by the officer [BZ(Ce)]
is transmitted with a small dephasing to the soldier (F) due to
a slightly slow non-commissioned officer (P). The other feed-
BZ(Ce) , the period for which is 45 s, transmits an excitatory
1
periodic UV signal to the oscillatory BZ(Ru), having an
intrinsic period of 15 s. After a few tens of seconds, a 3:1 phase
locking condition is reached as shown in Figure 1h and
simulated, for different periods of oscillations, in Figure 1i
forward network, FFN , is an upgrade of the g architecture,
2
because the three ANs work into three distinct cuvettes.
Specifically, two Orban reactions, Orb and Orb , character-
1
2
(
see also the Supporting Information).
ized by two distinct periods (due to their different chemical
composition; see Scheme 1) send chemiluminescent excita-
tory blue flashes to F (Figure 3c). F gives rise to a periodic
green emission that is a convolution of the two excitatory
signals, S and L, which stand for the short and long periods of
Orb and Orb , respectively (Figure 3d and the Supporting
The introduction of an inhibitory signal (indicated by
a flattened arrowhead), in the communication between 2ANs,
extends the possible sync conditions. In Figure 2, there are
two examples of 2ANs-2S communications in which the P-
type photochrome E (Figure 2b) receives both an excitatory
UV and an inhibitory visible signal. In configuration (1),
OscBZ(Ce) (Figure 2a) is crossed by the output of a Xe lamp
and transmits a periodic UV and continuous visible signal to
E. The two ANs synchronize almost in-phase (Figure 2d) or
in phase (Supporting Information, Figure S8) depending on
the relative intensities of the two signals. In configuration (2),
E receives both a continuous excitatory UV signal from
a lamp and a periodic inhibitory visible signal transmitted by
OscBZ(Ce-Fe) that shows pronounced oscillations in trans-
mittance in the blue-green region of the visible spectrum
1
2
Information, Figure S11). FFN is an implementation of the
2
two-masters–one-slave relationship and is powered only by
chemical energy and does not require an external lamp to be
sustained.
When there are feedback relationships, RNs are imple-
mented. Two examples of RNs are depicted in Figure 4. Both
RNs involve one OscBZ(Ce) and two photo-excitable ANs,
and both are hybrids of g and a architectures. In fact, BZ(Ce)2
is confined in one cuvette and communicates with the other
two ANs dissolved in the same acetonitrile solution (the
compartmentalization function of micelles is not needed in
these cases because the two excitable ANs are chemically
(
Figure 2c). When the transmittance of BZ(Ce-Fe) is at the
minimum, the coloration of E is at its maximum and vice
versa. The achievement of anti-phase synchronization is
confirmed by the simulations (Figure S9).
By introducing a third AN, we implement the simplest
networks, with either feed-forward or recurrent configura-
compatible). In RN , BZ(Ce) transmits an excitatory periodic
1
UV signal to F and E. E tends to produce a color, whereas F
fluoresces. F and E exert a mutual inhibitory action. The
photochromic transformation of E from the colorless A to the
Figure 2. Left: Spectral evolutions of BZ(Ce-Fe), E, and BZ(Ce) . Middle: Configurations of the two optical communications, whose outputs are
2
shown on right.
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
Figure 3. Schemes and outputs of FFN (left) and FFN (right).
1
2
colored B form inhibits the fluorescence of F because B has
a larger absorption coefficient in the UV than A (Figure 4a).
On the other hand, the green emission of F inhibits the
coloration of E because it induces the photo-bleaching of the
photochromic chromophore. Experimentally, we observe that
BZ(Ce) synchronizes in an anti-phase condition with E
other hand, the colored form of H exerts especially a positive
effect on E coloration because it protects E from the visible
signal (Figure 4d). The output of RN is shown in Figure 4e;
2
the total absorbance in the visible region synchronizes almost
in phase with the UV signal sent by BZ(Ce). The simulations
demonstrate that such a synchronization is established among
all the components of the network (see Figure 4 f) and is not
limited to the T-type P as it is the case of the absence of the
visible signal (Figure S19).
(
Figure 4b). In fact, when the excitatory UV radiation is
strong, the fluorescence is intense, and the coloration of E is
inhibited. On the other hand, when UV radiation intensity
decreases, the fluorescence diminishes, and the coloration
intensifies. The excitatory UV signal sent by BZ(Ce) is in
phase with the fluorescence of F (see Figure 4c obtained by
simulations), whereas the signal of absorbance of E [A(E)] is
This investigation demonstrates that UV and visible
radiation can be exploited as a code for communication
among oscillatory and photo-excitable chemical systems.
Such a code can exert both excitatory and inhibitory effects.
It has the power of inducing in-phase, out-of-phase, anti-
phase synchronizations and phase-locking between the
transmitter(s) and the receiver(s), as does the chemical code
of neurotransmitters within networks of real neurons. These
in an anti-phase condition with both I [BZ(Ce)] and IEM(F)
T
(
see also the Supporting Information, Figures S14 and S15). A
different type of synchronization is achieved in RN . OscBZ-
2
(
Ce) transmits a periodic UV signal and a continuous visible
[
13]
signal to two photochromic species: The P-type E and the T-
type H. E and H are in the same cuvette. The UV signal is
excitatory for both receivers, whereas the visible signal is
inhibitory only for E. The colored form of E exerts an
inhibitory screening effect on the coloration of H; on the
results are concordant with the embryonic idea of devel-
oping artificial neuron models communicating through light.
We are aware that in the field of neuromorphic engineering,
the exploitation of exhaustible, slow (note that the frequency
of spiking for our systems is of the order of tenths of hertz
4
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
ꢀ
4
ꢀ5
ꢀ5
Figure 4. Scheme and output of RN , wherein [F]=1.2ꢀ10 m and [E]=1.4ꢀ10 m, on the left, and RN , wherein [E]=5ꢀ10 m and
1
2
ꢀ5
[
H]=5ꢀ10 m, on the right. In (c) and (f), nI , nI and nA are the normalized transmitted, emitted intensities and absorbance, respectively.
T EM
versus the tens of hertz of real neurons), and bulky oscillatory
reactions have strong limitations compared to much faster
chemicals or electrons, will bring two great benefits. First, in
biological multicellular systems, which are complex out-of-
equilibrium soups of chemicals, the phenomena of long-range
coupling, mediated by diffusive chemicals, exist. Such long-
range coupling phenomena, not possible in the solid state,
enlarge the computing power of the whole system because
they originate collective oscillations and waves. Finally, the
encoding of information by UV/visible radiation will guaran-
tee very fast propagation of messages, easy tunability of their
[6]
(
reaching a switching frequency of gigahertz ) and easily
scalable traditional electronic components and memristors.
However, we believe that after miniaturizing our chemical
systems by micro-beads, microcapsules, and compartmental-
[12]
ization through micelles, liposomes, and micro-emulsions,
the use of a wetware (that is, solutions) rather than a hardware
solid state), and of electromagnetic radiation instead of
(
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!

Angewandte
Communications
Chemie
content and a wealthy code for futuristic opto-brain-like
computing machines.
[6] a) S. D. Ha, S. Ramanathan, J. Appl. Phys. 2011, 110, 071101;
b) J. J. Yang, D. B. Strukov, D. R. Stewart, Nat. Nanotechnol.
2013, 8, 13 – 24; c) A. Adamatzky, L. Chua, Memristor Networks,
Springer, New York, 2014.
[
[
[
7] a) P. L. Gentili, RSC Adv. 2013, 3, 25523 – 25549; b) G. Gruenert,
K. Gizynski, G. Escuela, B. Ibrahim, J. Gorecki, P. Dittrich, Int. J.
Neural Syst. 2015, 25, 1450032; c) N. Li, J. Delgado, H. O.
Gonzꢀlez-Ochoa, I. R. Epstein, S. Fraden, Phys. Chem. Chem.
Phys. 2014, 16, 10965 – 10978.
8] a) I. R. Epstein, J. A Pojman, An introduction to nonlinear
chemical dynamics: Oscillations, waves, patterns and chaos,
Oxford University Press, New York, 1998; b) P. Gray, S. K.
Scott, Chemical Oscillations and Instabilities. Non-Linear Chem-
ical Kinetics, Oxford University Press, USA, 1994.
Acknowledgements
Authors acknowledge the financial support by the University
of Perugia, Fondo Ricerca di Base 2014, D. D. n. 170, 23/12/
2
014.
Conflict of interest
9] a) E. M. Izhikevich, Dynamical Systems in Neuroscience, MIT
Press, Cambridge, 2007; b) M. I. Rabinovich, P. Varona, A. I.
Selverston, H. D. I. Abarbanel, Rev. Mod. Phys. 2006, 78, 1213 –
The authors declare no conflict of interest.
1
265.
Keywords: fluorescence · neural networks · oscillations ·
photochromism
[
[
10] M. Marek, I. Stuchl, Biophys. Chem. 1975, 3, 241 – 248.
11] a) M. F. Crowley, R. J. Field, J. Phys. Chem. 1986, 90, 1907 –
1915; b) W. Hohmann, M. Kraus, F. W. Schneider, J. Phys.
Chem. A 1999, 103, 7606 – 7611.
[
12] a) A. F. Taylor, M. R. Tinsley, K. Showalter, Phys. Chem. Chem.
Phys. 2015, 17, 20047 – 20055; b) V. Horvath, P. L. Gentili, V. K.
Vanag, I. R. Epstein, Angew. Chem. Int. Ed. 2012, 51, 6878 –
[
1] J. S. Albus, et al., Science 2007, 317, b1321.
[
2] a) G. Burr, et al., Adv. Phys. 2017, 2, 89 – 124; b) A. M. Shen, C.-
L. Chen, K. Kim, B. Cho, A. Tudor, Y. Chen, ACS Nano 2013, 7,
6
881; Angew. Chem. 2012, 124, 6984 – 6987.
[
13] F. Pina, M. J. Melo, M. Maestri, P. Passaniti, V. Balzani, J. Am.
Chem. Soc. 2000, 122, 4496 – 4498.
6117 – 6122.
[
3] C. Mead, Analog VLSI and Neural Systems, Addison-Wesley
Longman Publishing Co., Inc. Boston, MA, USA, 1989.
4] J. P. Donoghue, Neuron 2008, 60, 511 – 521.
5] a) R. F. Service, Science 2014, 345, 614 – 616; b) P. A. Merolla,
et al., Science 2014, 345, 668 – 673.
[
[
Manuscript received: March 3, 2017
Revised manuscript received: April 27, 2017
Version of record online: && &&, &&&&
6
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Nonequilibrium Processes
Some light conversation: UV/visible
radiation is a rich code for communicat-
ing excitatory and inhibitory signals
P. L. Gentili,* M. S. Giubila, R. Germani,
A. Romani, A. Nicoziani, A. Spalletti,
among oscillatory reactions and photo-
excitable systems, functioning as artificial
neuron models. Pairs or triads of such
chemical systems organized in both feed-
forward and recurrent networks syn-
chronize in phase, out of phase, in anti-
phase and phase-locking as real neurons
do through chemical neuro-transmitters.
B. M. Heron
&&&& — &&&&
Optical Communication among
Oscillatory Reactions and Photo-Excitable
Systems: UV and Visible Radiation Can
Synchronize Artificial Neuron Models
Angew. Chem. Int. Ed. 2017, 56, 1 – 7
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7
These are not the final page numbers!