1. Chapter 25 : Autocommunication 5/5/11
Lecture #25
Chapter 25 : Autocommunication
5/5/11

2. Key dates Final exam questions will be posted Friday by noon
Wiki projects are due Tuesday, May 10th at midnight
Last class on Tuesday
Final exam Thursday, May 12th 8 am

3. Today Echolocation basics Echolocation in bats
Open-site foragers
Gleaners
Hawkers and fishers
Echolocation in cetaceans

4. Autocommunication Echolocation = biosonar Electrolocation
Emit pulses of sound
Electrolocation
Determine environmental information about nearby objects
What is its size and shape?
What is it doing?
Where is it now and where might it be going?

5. Autocommunication Not technically communication
Shaped by same design factors such as…???

6. Autocommunication Not technically communication
Shaped by same design factors
Shape signals to adjust for
- Efficient emission
- Propagation and distortion
- Reception

7. Why is environmental signaling simple or absent much of time?
Do senders rarely reap benefits of communicating more complex information (benefits go to receivers)? OR
Is it too difficult and expensive to encode and extract information?

8. Autocommunication should represent perfect communication
Sender and receiver are the same individual - NO CONFLICT
Receiver should be ideal receiver - extract as much information as possible
Sender should pack as much information in as possible

9. Autocommunication should be perfect test case
So are autocommunication signals more complex than those between individuals? (no game)
Or are they just as simple because it is too difficult to encode and extract info?

10. Echolocation Simple echolocation More sophisticated echolocation
Oilbird Lesser tenrec Shrew
Oilbird:

11. Echolocation information
Environmental information:
Object location (range, angle)
Object identity (shape, texture composition)
Relative velocity and trajectory
May be trade offs btn object properties
Accuracy about one may require uncertainty in another

12. Detecting targets at distance
Outgoing signal must be loud enough to produce echo that is above ambient noise
Signal must be emitted enough of time that sound beam will strike target
Best if work at single frequency
All energy at that frequency
Tune receiver to that frequency

13. Echo intensity  1/distance4
Echo  1/distance2
Signal  1/distance2

14. Scattering summary Fig 2.6
Bigger objects, more scattering. Shorter wavelength, more scattering.
For 1 kHz, wavelength is 33 cm. So for objects < 5 cm, Rayleigh and objects>3.3 m, simple scattering. In between get complex scattering pattern.
Fig 2.6

15. Other factors Want frequency that can emit with highest intensity
Maximum reflection when wavelength < object
For 1cm insect, minimum freq is 34 kHz
For 10 cm fish, minimum freq is 3 kHz
But high frequencies absorbed by medium
So optimum is in between
Frequency = c / wavelength = 340 m/s / .01 m = 34 kHz

16. Echolocation signals
Assumes bat signal of 110 dB at 60 kHz directed at 1.9 cm target located 3 m away
If sound goes from 0.1 m to 3 m then spreading loss is 10 log ((.1/3)^2)) = -30 dB but looks like only -15 dB on returning - I think absorption and spreading loss are switched
Outgoing absorption is about 5 dB?? But returning absorption is more like 30 dB
Lose 20 dB in echo (factor of 100

17. Echolocation signals - optimal frequency is balance of production and propagation
Assumes bat signal of 110 dB at 60 kHz directed at 1.9 cm target located 3 m away
If sound goes from 0.1 m to 3 m then spreading loss is 10 log ((.1/3)^2)) = -30 dB but looks like only -15 dB on returning - I think absorption and spreading loss are switched
Outgoing absorption is about 5 dB?? But returning absorption is more like 30 dB
Lose 20 dB in echo (factor of 100

18. Pulse duration Send out pulse of sound
Need to leave silence to hear echo
Time between pulses (trep) = 1 / frequency
Duty cycle = fraction of time are calling per pulse
= tpulse / trep
tpulse
trep

19. Time for echo to return Sound velocity = distance / time
Return time = distance / velocity
Distance = Velocity * time
Distance to object

20. Time for echo to return
Signal
Echo
techo

21. Minimum distance probed determined by pulse width
Signal
tpulse
Echo
Have to wait for outgoing pulse to end before can hear echo
techo =
tpulse

22. Minimum distance probed determined by pulse width
Signal
tpulse
Echo
Pulse width
Distance (m)
0.1 s
17 m
0.01 s
1.7 m
0.001 s
0.17 m
techo =
tpulse

23. Maximum distance probed determined by pulse repetition
Signal
tpulse
techo = trep
Echo
Want to receive echo before next pulse begins

24. Maximum distance probed determined by pulse repetition
Signal
tpulse
Echo
techo = trep
Freq (Hz)
Pulse rep
Distance 20
0.05
8.5 m
100
0.01
1.7
500
0.002
0.34

25. Determining target distance
Can use echo amplitude
Also affected by size and shape of target
Can use delay time, techo
Possible that moth only reflect sound when its wings are perpendicular to sound (top and bottom of flap)
So only get part of pulse returned
If label pulse with frequency, will know which part of pulse is reflected

26. Uncertainties
Constant frequency pulse has errors
tactual
tmeasured

27. Frequency modulation is more accurate
Can compare correlated part of the pulse
tactual
Can decrease range error from
2.7 to 0.16 cm
tmeasured

28. Doppler demonstration#2
Doppler demonstration

29. Doppler shift
Perceived frequency depends on relative velocity of predator and prey
If moving towards, frequency is higher (minus sign)
If moving away, frequency is lower (plus sign)

30. Doppler shifts
As bat flies towards moth, echo will be doppler shifted higher
Also will arrive sooner than if bat is stationary
Can linearly modulate emitted pulse
Frequency Period Inverse
frequency
So modulation of pulse period (making it shorter as go) produces most accurate and predictable shift (most like outgoing pulse) which is good for comparisons and makes for Doppler tolerant signal

31. Determining target angle
If focus outgoing beam to narrow angle, then know at what target angle get a response
High resolution
May miss insect
Angle may vary with frequency
Can also use directional sensitivity of ears
Arrival time delay

32. Emitted beam has angular variation Compensate with angular response of ears!!
Emitted beam Ear sensitivity Combination

33. Determining target properties
Distinguish echoes from background (clutter) and target prey
Echo amplitude related to target size
-- if also know target range (time delay)
All aerial targets will reflect
No info on what target is made of
Most aquatic targets will absorb and may resonate - can use to discriminate what target is made of

34. Determining target velocity and trajectory
Hard to do with frequency modulated pulse
Which outgoing frequency matches which incoming frequency??
Using constant frequency signals, can get better estimate of Doppler shifts
Get more cycles of sound on target

35. Tradeoff Frequency modulation Constant frequency
Best for estimating range
Constant frequency
Best for estimating velocity
Tough to estimate both range and velocity at same time
Decreased error for one increases error for other
Animals must choose which to optimize

36. Echolocation in bats 800 species of bats Megachiroptera
Eat fruit, nector or flowers
Rely on nocturnal vision (one rudimentary echolocator)
Microchiroptera
Feed on insects, small verts, blood or fruit/flowers
Use echolocation to forage and navigate

37. Open-site foragers : insectivores
Not much clutter to worry about
Emit broad beam of sound through mouth (60-90 deg)
Loud intensities dB SPL
As intense as can be w/o collapse blood vessels
Three phases of foraging
Search
Approach
Buzz

39. Decreases pulse width as get closer to target
Always stay below echo overlap pulse width
When very close, stop echolocating (4-10 cm)

40. Open-site foragers

41. Range Pipistrelle is small bat Larger bats use lower frequencies
Use kHz to search
Can detect prey > 0.2 mm
Maximum range m
Larger bats use lower frequencies
Either to find larger prey
Or to find prey at larger distance since not as agile

42. Higher frequency is used at shorter range, requires shorter pulse duration
Higher frequencies are more quickly attenuated and so they sample closer to the bat. At these shorter distances, echos reach bat more quickly and so pulse duration has to be shorter to avoid pulse - echo overlap.
Px are the different Pipistrelle species

43. Bats are pretty good at this
Time to catch an insect = s
Catch insect every 4 s
Capture efficiency = 30-40% for moths
= 60-70% for mosquitos

44. Bat - Gleaners Forage from close to surfaces
Foliage, barns
Lots of clutter from larger reflecting surface
To prevent being overwhelmed use low sound intensity
Whispering bats
Narrow beam emitted from nose

45. Insectivorous bats Mouth emission Frugivore Nose emission Predators
Spiders, lizards Vampire doves, parrots
Fishing hawks insects gleans insect

46. Detect moving target against fixed background
Need to determine angle of moving target
Gleaners have high angular accuracy 1.5-2º
Can also perceive fluttering motion in reflected spectrum
Can also detect sounds from prey and odors

47. Hawkers and fishers
Hang in one location scanning for prey and then fly in to capture
Patrol close to vegetation and capture moving insects
Fishers detect disturbance on water surface
Need to then predict where fish will be when go to catch it (sonar does not enter water)

48. Fishing bats and finding water

49. Hawking by Rhinolophus
Use high duty cycle constant freq pulses with FM sweeps on end
Use very high freq so can detect small targets
Doppler shifted echoes from prey movement
Want wavelength about size of animal 83kHz = 4 mm object
If insect moves 1 m/s then freq shift of 484 Hz vs 83kHz

50. Cochlea has hair cells supported by basilar membrane
Basilar membrane responds to high, med and low freq as move down the cochlea - tonotopic map

51. CF bat ears Normal mammalian cochlea has tonotopic map
CF bat cochlea has a “fovea”
Region where tonotopic map is expanded …

52. Bats are close to ideal receivers
Design ideal signals
FM sweeps which are Doppler-tolerant
Likely use comparison of phase shifts in echoes vs emitted calls to further improve ranging (cross correlation)
Gives minimal error possible
0.007 mm
May help with target identification

53. Why is environmental signaling simple or absent much of time?
Do senders rarely reap benefits of communicating more complex information (benefits go to receivers)? OR
Is it too difficult and expensive to encode and extract information?
Bats use most complex signals possible

54. Echolocation in cetaceans
Predatory species echolocate
Toothed whales
Porpoises
Challenges of aquatic habitat
Velocity 4.4x greater; Wavelengths 4.4x longer
Need higher frequencies for same sized target (vs air)
But since larger animals, targets are larger so it’s a wash

55. Other constraints
Faster sound velocities require shorter pulses to prevent pulse - echo overlap ms
This is even shorter than needed so don’t have to shorten as get closer to an object as bats do
Low attenuation so good signal propagation
Dolphin can detect 2 cm target at 73m
Bat needs to be 2-4 m to detect it

56. How do cetaceans create pulses?
Porpoise emits through head and detects through jaw
Porpoise pulse and power spectrum
Dolphin detection is only 6-8dB worse than an ideal receiver Minimal range error < 1 cm

57. Cetacean pulse production
Sperm whale breathes through left nostril and makes pulses with right nostril

58. Conclusion Both bats and cetaceans are close to ideal receivers
So rest of animals could send and extract more information than they do
Likely conflicts of interest
Within bats, they do not share their environmental info with other bats
Have own personal frequencies

59. One exception to all of this
Humans
We share all kinds of environmental information (and other information) with each other
Perhaps we will find other animals do better than we think: parrots, corvids, elephants, higher primates and cetaceans

60. Congratulations!! I can’t believe we read the WHOLE book
There is still more to learn about animal communication