Abstract
Stomatopods, or mantis shrimps, are a group of crustaceans that possess an elaboratevisual system which requires the eye to be compartmentalised to process different visual
modalities. This study intended to investigate whether visual querying of unknown objects is
processed through the chromatic midband region or the achromatic hemispheres, by
producing an action spectrum that could be compared to other action spectra retrieved from
the hemispheres of other stomatopod species. This spectral sensitivity curve would display
absolute sensitivities as opposed to normalised ones, something lacking from the current
literature. 12 Odontodactylus scyllarus (family: Odontodactylidae; superfamily:
Gonodactyloidea) individuals were trained through operant conditioning to emerge from their
burrows for a food reward upon detecting a flashing LED stimulus. Individuals were trained
by a white stimulus with a simultaneously deployed food reward, before being assessed on
their responses to six different wavelength stimuli (450, 500, 550, 600, 650, and 700 nm)
where the reward was deployed only after a successful response. An automated system that
would train and test each individual stomatopod with minimal levels of researcher labour was
designed and constructed. The experimental method was continuously modified and finetuned throughout the research in an attempt to create an optimal automated system and
procedure for future use.
Training became steadily more effective and efficient with the method modifications,
however only four individuals were deemed sufficiently conditioned to attempt wavelength
trials, partly due to time constraints. The most successful trial design utilised a trial ratio
system, large food reward, choice window of 30s, control probability of 0.25, and neutraldensity (ND) filters applied to each LED stimulus. Stomatopods could likely detect all six
wavelengths but a high degree of noise in the results meant a detection threshold for each
wavelength could not be extracted. Without this information, creating a spectral sensitivity
curve was not possible and thus neither was revealing the ocular region responsible for it.
With suggested improvements, it is most likely this automated system would successfully
reveal this information. A potential spectral sensitivity curve based on physiological and
ecological findings is suggested and perhaps the use of hemispheres and midband together
in querying objects.
| Date of Award | 21 Jun 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Martin J How (Supervisor) & Nicholas W Roberts (Supervisor) |